upstream/haproxy
1
0
Fork 0
mirror of https://github.com/haproxy/haproxy.git synced 2026-04-15 21:59:41 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions 3
haproxy/src/dumpstats.c

4152 lines
127 KiB
C
Raw Normal View History

[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/*
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
* Functions dedicated to statistics output and the stats socket
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
*
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
* Copyright 2000-2010 Willy Tarreau <w@1wt.eu>
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
* Copyright 2007-2009 Krzysztof Piotr Oledzki <ole@ans.pl>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
*/
#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
[MINOR] Allow listing of stick table by key
2011-06-15 02:18:47 -04:00
#include <stdbool.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
#include <pwd.h>
#include <grp.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/types.h>
[MEDIUM] modularize the global "stats" keyword configuration parser The "stats" keyword already relied on an external parser, let's make use of the new keyword registration mechanism.
2008-07-09 14:12:41 -04:00
#include <common/cfgparse.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <common/compat.h>
#include <common/config.h>
#include <common/debug.h>
#include <common/memory.h>
#include <common/mini-clist.h>
#include <common/standard.h>
[MAJOR] convert all expiration timers from timeval to ticks This is the first attempt at moving all internal parts from using struct timeval to integer ticks. Those provides simpler and faster code due to simplified operations, and this change also saved about 64 bytes per session. A new header file has been added : include/common/ticks.h. It is possible that some functions should finally not be inlined because they're used quite a lot (eg: tick_first, tick_add_ifset and tick_is_expired). More measurements are required in order to decide whether this is interesting or not. Some function and variable names are still subject to change for a better overall logics.
2008-07-06 18:09:58 -04:00
#include <common/ticks.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <common/time.h>
#include <common/uri_auth.h>
#include <common/version.h>
#include <types/global.h>
#include <proto/backend.h>
REORG: buffers: split buffers into chunk,buffer,channel Many parts of the channel definition still make use of the "buffer" word.
2012-08-24 13:22:53 -04:00
#include <proto/channel.h>
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
#include <proto/checks.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <proto/dumpstats.h>
#include <proto/fd.h>
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
#include <proto/freq_ctr.h>
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
#include <proto/log.h>
[STATS] report pipe usage in the statistics Pipe usage is reported in info and web stats including maxpipes, pipes_free, and pipes_used.
2009-01-25 08:02:00 -05:00
#include <proto/pipe.h>
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
#include <proto/protocols.h>
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
#include <proto/proto_uxst.h>
[MEDIUM] make the global stats socket part of a frontend Creating a frontend for the global stats socket will help merge unix sockets management with the other socket management. Since frontends are huge structs, we only allocate it if required.
2009-08-16 11:41:45 -04:00
#include <proto/proxy.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
#include <proto/session.h>
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
#include <proto/server.h>
REORG: sock_raw: rename the files raw_sock* The "raw_sock" prefix will be more convenient for naming functions as it will be prefixed with the data layer and suffixed with the data direction. So let's rename the files now to avoid any further confusion. The #include directive was also removed from a number of files which do not need it anymore.
2012-08-20 11:01:35 -04:00
#include <proto/raw_sock.h>
[MINOR] replace client_retnclose() with stream_int_retnclose() This makes more sense to return a message to a stream interface than to a session. senddata.{c,h} have been removed.
2008-11-30 13:48:07 -05:00
#include <proto/stream_interface.h>
[OPTIM] task: don't unlink a task from a wait queue when waking it up In many situations, we wake a task on an I/O event, then queue it exactly where it was. This is a real waste because we delete/insert tasks into the wait queue for nothing. The only reason for this is that there was only one tree node in the task struct. By adding another tree node, we can have one tree for the timers (wait queue) and one tree for the priority (run queue). That way, we can have a task both in the run queue and wait queue at the same time. The wait queue now really holds timers, which is what it was designed for. The net gain is at least 1 delete/insert cycle per session, and up to 2-3 depending on the workload, since we save one cycle each time the expiration date is not changed during a wake up.
2009-03-07 11:25:21 -05:00
#include <proto/task.h>
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_raw_to_buffer(struct stream_interface *si);
static int stats_dump_full_sess_to_buffer(struct stream_interface *si);
static int stats_dump_sess_to_buffer(struct stream_interface *si);
static int stats_dump_errors_to_buffer(struct stream_interface *si);
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
static int stats_table_request(struct stream_interface *si, bool show);
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_proxy(struct stream_interface *si, struct proxy *px, struct uri_auth *uri);
static int stats_dump_http(struct stream_interface *si, struct uri_auth *uri);
static struct si_applet cli_applet;
static const char stats_sock_usage_msg[] =
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
"Unknown command. Please enter one of the following commands only :\n"
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
" clear counters : clear max statistics counters (add 'all' for all counters)\n"
[MEDIUM] stats: add "clear table <name> key <value>" to clear table entries This feature will be required at some point, when the stick tables are used to enforce security measures. For instance, some visitors may be incorrectly flagged as abusers and would ask the site admins to remove their entry from the table.
2010-07-13 07:48:00 -04:00
" clear table : remove an entry from a table\n"
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
" help : this message\n"
" prompt : toggle interactive mode with prompt\n"
" quit : disconnect\n"
" show info : report information about the running process\n"
" show stat : report counters for each proxy and server\n"
" show errors : report last request and response errors for each proxy\n"
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
" show sess [id] : report the list of current sessions or dump this session\n"
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
" show table [id]: report table usage stats or dump this table's contents\n"
[MEDIUM] stats: add the "get weight" command to report a server's weight This command is unrestricted and simply reports a server's weight, as well as the weight initially configured, which at the moment are the sames.
2009-10-10 12:37:29 -04:00
" get weight : report a server's current weight\n"
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
" set weight : change a server's weight\n"
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
" set table [id] : update or create a table entry's data\n"
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
" set timeout : change a timeout setting\n"
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
" set maxconn : change a maxconn setting\n"
[MEDIUM] stats: add the ability to adjust the global maxconnrate Using "set rate-limit connections global <xxx>" on the CLI, we can now adjust the per-process connection rate limiting (equal to global.maxconnrate).
2011-09-07 10:13:44 -04:00
" set rate-limit : change a rate limiting value\n"
[MEDIUM] stats: offer the possibility to kill a session from the CLI It's now possible to issue "shutdown session 0xXXXXXXXX" and have this session immediately killed. Useful for long-running fantoms.
2011-09-07 17:21:03 -04:00
" disable : put a server or frontend in maintenance mode\n"
" enable : re-enable a server or frontend which is in maintenance mode\n"
" shutdown : kill a session or a frontend (eg:to release listening ports)\n"
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
"";
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static const char stats_permission_denied_msg[] =
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
"Permission denied\n"
"";
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
/* data transmission states for the stats responses */
enum {
STAT_ST_INIT = 0,
STAT_ST_HEAD,
STAT_ST_INFO,
STAT_ST_LIST,
STAT_ST_END,
STAT_ST_FIN,
};
/* data transmission states for the stats responses inside a proxy */
enum {
STAT_PX_ST_INIT = 0,
STAT_PX_ST_TH,
STAT_PX_ST_FE,
STAT_PX_ST_LI,
STAT_PX_ST_SV,
STAT_PX_ST_BE,
STAT_PX_ST_END,
STAT_PX_ST_FIN,
};
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
extern const char *stat_status_codes[];
[MEDIUM] stats: rely on the standard session_accept() function The stats' accept() function is now ridiculously small. It could even be reduced by moving some parts to the common accept code.
2010-06-01 12:03:19 -04:00
/* This function is called from the session-level accept() in order to instanciate
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
* a new stats socket. It returns a positive value upon success, 0 if the connection
* needs to be closed and ignored, or a negative value upon critical failure.
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_accept(struct session *s)
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
{
[MEDIUM] stats: rely on the standard session_accept() function The stats' accept() function is now ridiculously small. It could even be reduced by moving some parts to the common accept code.
2010-06-01 12:03:19 -04:00
/* we have a dedicated I/O handler for the stats */
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
stream_int_register_handler(&s->si[1], &cli_applet);
[BUG] log: retrieve the target from the session, not the SI Since we now have the copy of the target in the session, use it instead of relying on the SI for it. The SI drops the target upon unregister() so applets such as stats were logged as "NOSRV".
2011-03-27 13:53:06 -04:00
copy_target(&s->target, &s->si[1].target); // for logging only
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
s->si[1].conn.data_ctx = s;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
s->si[1].applet.st1 = 0;
s->si[1].applet.st0 = STAT_CLI_INIT;
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
[MEDIUM] stats: rely on the standard session_accept() function The stats' accept() function is now ridiculously small. It could even be reduced by moving some parts to the common accept code.
2010-06-01 12:03:19 -04:00
tv_zero(&s->logs.tv_request);
s->logs.t_queue = 0;
s->logs.t_connect = 0;
s->logs.t_data = 0;
s->logs.t_close = 0;
s->logs.bytes_in = s->logs.bytes_out = 0;
s->logs.prx_queue_size = 0; /* we get the number of pending conns before us */
s->logs.srv_queue_size = 0; /* we will get this number soon */
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
s->req->flags |= BF_READ_DONTWAIT; /* we plan to read small requests */
[MEDIUM] stats: rely on the standard session_accept() function The stats' accept() function is now ridiculously small. It could even be reduced by moving some parts to the common accept code.
2010-06-01 12:03:19 -04:00
if (s->listener->timeout) {
s->req->rto = *s->listener->timeout;
s->rep->wto = *s->listener->timeout;
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
}
return 1;
}
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
/* allocate a new stats frontend named <name>, and return it
* (or NULL in case of lack of memory).
*/
static struct proxy *alloc_stats_fe(const char *name)
{
struct proxy *fe;
fe = (struct proxy *)calloc(1, sizeof(struct proxy));
if (!fe)
return NULL;
[BUG] proxy: stats frontend and peers were missing many initializers This was revealed with one of the very latest patches which caused the listener_queue not to be initialized on the stats socket frontend. And in fact a number of other ones were missing too. This is getting so boring that now we'll always make use of the same function to initialize any proxy. Doing so has even saved about 500 bytes on the binary due to the avoided code redundancy. No backport is needed.
2011-07-28 19:49:03 -04:00
init_new_proxy(fe);
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
fe->last_change = now.tv_sec;
fe->id = strdup("GLOBAL");
fe->cap = PR_CAP_FE;
[MEDIUM] stats: disable complex socket reservation for stats socket The way the unix socket is initialized is awkward. Some of the settings are put in the sockets itself, other ones in the backend. And more importantly the global.maxsock value is adjusted so that the stats socket evades the global maxconn value. This complexifies maxsock computations for nothing, since the stats socket is not supposed to receive hundreds of concurrent connections when the global maxconn is very low. What is needed however is to ensure that there are always connections left for the stats socket even when traffic sockets are saturated, but this guarantee is not offered anymore by current code. So as of now, the stats socket is subject to the global maxconn limitation just as any other socket until a reservation mechanism is implemented.
2011-09-07 06:13:34 -04:00
fe->maxconn = 10; /* default to 10 concurrent connections */
fe->timeout.client = MS_TO_TICKS(10000); /* default timeout of 10 seconds */
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
return fe;
}
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
/* This function parses a "stats" statement in the "global" section. It returns
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
* -1 if there is any error, otherwise zero. If it returns -1, it will write an
* error message into the <err> buffer which will be preallocated. The trailing
* '\n' must not be written. The function must be called with <args> pointing to
* the first word after "stats".
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
*/
[MEDIUM] modularize the global "stats" keyword configuration parser The "stats" keyword already relied on an external parser, let's make use of the new keyword registration mechanism.
2008-07-09 14:12:41 -04:00
static int stats_parse_global(char **args, int section_type, struct proxy *curpx,
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
struct proxy *defpx, char **err)
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
{
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
if (!strcmp(args[1], "socket")) {
[MINOR] cfgparse: Check whether the path given for the stats socket actually fits into the sockaddr_un structure to avoid truncation. while working further on the changes to allow for dynamic adding/removing of backend servers we noticed a potential problem: the path given for the 'stats socket' global option may get truncated when copying it into the sockaddr_un.sun_path field. Attached patch checks the length, and reports an error if truncation would happen. This issue was noticed by Joerg Sonnenberger <joerg@NetBSD.org>.
2011-01-19 14:29:32 -05:00
struct sockaddr_un *su;
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
int cur_arg;
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
if (*args[2] == 0) {
memprintf(err, "'%s %s' in global section expects a path to a UNIX socket", args[0], args[1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
if (global.stats_sock.state != LI_NEW) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' already specified in global section", args[0], args[1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
su = str2sun(args[2]);
[MINOR] cfgparse: Check whether the path given for the stats socket actually fits into the sockaddr_un structure to avoid truncation. while working further on the changes to allow for dynamic adding/removing of backend servers we noticed a potential problem: the path given for the 'stats socket' global option may get truncated when copying it into the sockaddr_un.sun_path field. Attached patch checks the length, and reports an error if truncation would happen. This issue was noticed by Joerg Sonnenberger <joerg@NetBSD.org>.
2011-01-19 14:29:32 -05:00
if (!su) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : path would require truncation", args[0], args[1]);
[MINOR] cfgparse: Check whether the path given for the stats socket actually fits into the sockaddr_un structure to avoid truncation. while working further on the changes to allow for dynamic adding/removing of backend servers we noticed a potential problem: the path given for the 'stats socket' global option may get truncated when copying it into the sockaddr_un.sun_path field. Attached patch checks the length, and reports an error if truncation would happen. This issue was noticed by Joerg Sonnenberger <joerg@NetBSD.org>.
2011-01-19 14:29:32 -05:00
return -1;
}
memcpy(&global.stats_sock.addr, su, sizeof(struct sockaddr_un)); // guaranteed to fit
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
[MEDIUM] make the global stats socket part of a frontend Creating a frontend for the global stats socket will help merge unix sockets management with the other socket management. Since frontends are huge structs, we only allocate it if required.
2009-08-16 11:41:45 -04:00
if (!global.stats_fe) {
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
if ((global.stats_fe = alloc_stats_fe("GLOBAL")) == NULL) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : out of memory trying to allocate a frontend", args[0], args[1]);
[MEDIUM] make the global stats socket part of a frontend Creating a frontend for the global stats socket will help merge unix sockets management with the other socket management. Since frontends are huge structs, we only allocate it if required.
2009-08-16 11:41:45 -04:00
return -1;
}
}
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
global.stats_sock.state = LI_INIT;
[MEDIUM] don't limit peers nor stats socket to maxconn nor maxconnrate The peers and the stats socket are control sockets, they must not be limited by traffic rules.
2011-09-07 12:00:47 -04:00
global.stats_sock.options = LI_O_UNLIMITED;
[MEDIUM] stats: rely on the standard session_accept() function The stats' accept() function is now ridiculously small. It could even be reduced by moving some parts to the common accept code.
2010-06-01 12:03:19 -04:00
global.stats_sock.accept = session_accept;
global.stats_fe->accept = stats_accept;
[MEDIUM] make the unix stats sockets use the generic session handler process_session() is now ready to handle unix stats sockets. This first step works and old code has not been removed. A cleanup is required. The stats handler is not unix socket-centric anymore and should move to dumpstats.c.
2009-08-16 12:51:29 -04:00
global.stats_sock.handler = process_session;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
global.stats_sock.analysers = 0;
[MINOR] move the initial task's nice value to the listener Since the listener is the one indicating what analyser and session handlers to call, it makes sense that it also sets the task's nice value. This also helps getting rid of the last trace of the stats in the proto_uxst file.
2009-08-16 13:12:36 -04:00
global.stats_sock.nice = -64; /* we want to boost priority for local stats */
[MEDIUM] separate protocol-level accept() from the frontend's For a long time we had two large accept() functions, one for TCP sockets instanciating proxies, and another one for UNIX sockets instanciating the stats interface. A lot of code was duplicated and both did not work exactly the same way. Now we have a stream_sock layer accept() called for either TCP or UNIX sockets, and this function calls the frontend-specific accept() function which does the rest of the frontend-specific initialisation. Some code is still duplicated (session & task allocation, stream interface initialization), and might benefit from having an intermediate session-level accept() callback to perform such initializations. Still there are some minor differences that need to be addressed first. For instance, the monitor nets should only be checked for proxies and not for other connection templates. Last, we renamed l->private as l->frontend. The "private" pointer in the listener is only used to store a frontend, so let's rename it to eliminate this ambiguity. When we later support detached listeners (eg: FTP), we'll add another field to avoid the confusion.
2010-05-28 12:46:57 -04:00
global.stats_sock.frontend = global.stats_fe;
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
global.stats_sock.perm.ux.level = ACCESS_LVL_OPER; /* default access level */
[MEDIUM] stats: disable complex socket reservation for stats socket The way the unix socket is initialized is awkward. Some of the settings are put in the sockets itself, other ones in the backend. And more importantly the global.maxsock value is adjusted so that the stats socket evades the global maxconn value. This complexifies maxsock computations for nothing, since the stats socket is not supposed to receive hundreds of concurrent connections when the global maxconn is very low. What is needed however is to ensure that there are always connections left for the stats socket even when traffic sockets are saturated, but this guarantee is not offered anymore by current code. So as of now, the stats socket is subject to the global maxconn limitation just as any other socket until a reservation mechanism is implemented.
2011-09-07 06:13:34 -04:00
global.stats_sock.maxconn = global.stats_fe->maxconn;
[MEDIUM] make the global stats socket part of a frontend Creating a frontend for the global stats socket will help merge unix sockets management with the other socket management. Since frontends are huge structs, we only allocate it if required.
2009-08-16 11:41:45 -04:00
global.stats_sock.timeout = &global.stats_fe->timeout.client;
global.stats_sock.next = global.stats_fe->listen;
global.stats_fe->listen = &global.stats_sock;
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
cur_arg = 3;
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
while (*args[cur_arg]) {
if (!strcmp(args[cur_arg], "uid")) {
global.stats_sock.perm.ux.uid = atol(args[cur_arg + 1]);
cur_arg += 2;
}
else if (!strcmp(args[cur_arg], "gid")) {
global.stats_sock.perm.ux.gid = atol(args[cur_arg + 1]);
cur_arg += 2;
}
else if (!strcmp(args[cur_arg], "mode")) {
global.stats_sock.perm.ux.mode = strtol(args[cur_arg + 1], NULL, 8);
cur_arg += 2;
}
else if (!strcmp(args[cur_arg], "user")) {
struct passwd *user;
user = getpwnam(args[cur_arg + 1]);
if (!user) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : unknown user '%s'", args[0], args[1], args[cur_arg + 1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
global.stats_sock.perm.ux.uid = user->pw_uid;
cur_arg += 2;
}
else if (!strcmp(args[cur_arg], "group")) {
struct group *group;
group = getgrnam(args[cur_arg + 1]);
if (!group) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : unknown group '%s'", args[0], args[1], args[cur_arg + 1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
global.stats_sock.perm.ux.gid = group->gr_gid;
cur_arg += 2;
}
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
else if (!strcmp(args[cur_arg], "level")) {
if (!strcmp(args[cur_arg+1], "user"))
global.stats_sock.perm.ux.level = ACCESS_LVL_USER;
else if (!strcmp(args[cur_arg+1], "operator"))
global.stats_sock.perm.ux.level = ACCESS_LVL_OPER;
else if (!strcmp(args[cur_arg+1], "admin"))
global.stats_sock.perm.ux.level = ACCESS_LVL_ADMIN;
else {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : '%s' only supports 'user', 'operator', and 'admin' (got '%s')",
args[0], args[1], args[cur_arg], args[cur_arg+1]);
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
return -1;
}
cur_arg += 2;
}
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
else {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' only supports 'user', 'uid', 'group', 'gid', 'level', and 'mode' (got '%s')",
args[0], args[1], args[cur_arg]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
}
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
CLEANUP: connection: split sock_ops into data_ops, app_cp and si_ops Some parts of the sock_ops structure were only used by the stream interface and have been moved into si_ops. Some of them were callbacks to the stream interface from the connection and have been moved into app_cp as they're the application seen from the connection (later, health-checks will need to use them). The rest has moved to data_ops. Normally at this point the connection could live without knowing about stream interfaces at all.
2012-08-24 12:12:41 -04:00
global.stats_sock.data = &raw_sock;
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
uxst_add_listener(&global.stats_sock);
global.maxsock++;
}
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
else if (!strcmp(args[1], "timeout")) {
[MEDIUM] add support for time units in the configuration It is not always handy to manipulate large values exprimed in milliseconds for timeouts. Also, some values are entered in seconds (such as the stats refresh interval). This patch adds support for time units. It knows about 'us', 'ms', 's', 'm', 'h', and 'd'. It automatically converts each value into the caller's expected unit. Unit-less values are still passed unchanged. The unit must be passed as a suffix to the number. For instance: clitimeout 15m If any character is not understood, an error is returned.
2007-12-02 16:15:14 -05:00
unsigned timeout;
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
const char *res = parse_time_err(args[2], &timeout, TIME_UNIT_MS);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
[MEDIUM] add support for time units in the configuration It is not always handy to manipulate large values exprimed in milliseconds for timeouts. Also, some values are entered in seconds (such as the stats refresh interval). This patch adds support for time units. It knows about 'us', 'ms', 's', 'm', 'h', and 'd'. It automatically converts each value into the caller's expected unit. Unit-less values are still passed unchanged. The unit must be passed as a suffix to the number. For instance: clitimeout 15m If any character is not understood, an error is returned.
2007-12-02 16:15:14 -05:00
if (res) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : unexpected character '%c'", args[0], args[1], *res);
[MEDIUM] add support for time units in the configuration It is not always handy to manipulate large values exprimed in milliseconds for timeouts. Also, some values are entered in seconds (such as the stats refresh interval). This patch adds support for time units. It knows about 'us', 'ms', 's', 'm', 'h', and 'd'. It automatically converts each value into the caller's expected unit. Unit-less values are still passed unchanged. The unit must be passed as a suffix to the number. For instance: clitimeout 15m If any character is not understood, an error is returned.
2007-12-02 16:15:14 -05:00
return -1;
}
if (!timeout) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' expects a positive value", args[0], args[1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
if (!global.stats_fe) {
if ((global.stats_fe = alloc_stats_fe("GLOBAL")) == NULL) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : out of memory trying to allocate a frontend", args[0], args[1]);
[BUG] stats: global stats timeout may be specified before stats socket. If the global stats timeout statement was found before the stats socket (or without), the parser would crash because the stats frontend was not initialized. Now we have an allocation function which solves the issue. This bug was introduced with 1.4 so it does not need backporting. (was commit 1c5819d2498ae3643c3880507847f948a53d2773 in 1.4)
2010-08-17 15:48:17 -04:00
return -1;
}
}
[MEDIUM] make the global stats socket part of a frontend Creating a frontend for the global stats socket will help merge unix sockets management with the other socket management. Since frontends are huge structs, we only allocate it if required.
2009-08-16 11:41:45 -04:00
global.stats_fe->timeout.client = MS_TO_TICKS(timeout);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
}
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
else if (!strcmp(args[1], "maxconn")) {
int maxconn = atol(args[2]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
if (maxconn <= 0) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' expects a positive value", args[0], args[1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
[MEDIUM] stats: disable complex socket reservation for stats socket The way the unix socket is initialized is awkward. Some of the settings are put in the sockets itself, other ones in the backend. And more importantly the global.maxsock value is adjusted so that the stats socket evades the global maxconn value. This complexifies maxsock computations for nothing, since the stats socket is not supposed to receive hundreds of concurrent connections when the global maxconn is very low. What is needed however is to ensure that there are always connections left for the stats socket even when traffic sockets are saturated, but this guarantee is not offered anymore by current code. So as of now, the stats socket is subject to the global maxconn limitation just as any other socket until a reservation mechanism is implemented.
2011-09-07 06:13:34 -04:00
if (!global.stats_fe) {
if ((global.stats_fe = alloc_stats_fe("GLOBAL")) == NULL) {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s %s' : out of memory trying to allocate a frontend", args[0], args[1]);
[MEDIUM] stats: disable complex socket reservation for stats socket The way the unix socket is initialized is awkward. Some of the settings are put in the sockets itself, other ones in the backend. And more importantly the global.maxsock value is adjusted so that the stats socket evades the global maxconn value. This complexifies maxsock computations for nothing, since the stats socket is not supposed to receive hundreds of concurrent connections when the global maxconn is very low. What is needed however is to ensure that there are always connections left for the stats socket even when traffic sockets are saturated, but this guarantee is not offered anymore by current code. So as of now, the stats socket is subject to the global maxconn limitation just as any other socket until a reservation mechanism is implemented.
2011-09-07 06:13:34 -04:00
return -1;
}
}
global.stats_fe->maxconn = maxconn;
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
}
else {
MEDIUM: cfgparse: use the new error reporting framework for remaining cfg_keywords All keywords registered using a cfg_kw_list now make use of the new error reporting framework. This allows easier and more precise error reporting without having to deal with complex buffer allocation issues.
2012-05-08 13:47:01 -04:00
memprintf(err, "'%s' only supports 'socket', 'maxconn' and 'timeout' (got '%s')", args[0], args[1]);
[MEDIUM] introduce the "stats" keyword in global section Removed old unused MODE_LOG and MODE_STATS, and replaced the "stats" keyword in the global section. The new "stats" keyword in the global section is used to create a UNIX socket on which the statistics will be accessed. The client must issue a "show stat\n" command in order to get a CSV-formated output similar to the output on the HTTP socket in CSV mode.
2007-10-18 07:53:22 -04:00
return -1;
}
return 0;
}
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int print_csv_header(struct chunk *msg)
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
{
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
return chunk_printf(msg,
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"# pxname,svname,"
"qcur,qmax,"
"scur,smax,slim,stot,"
"bin,bout,"
"dreq,dresp,"
"ereq,econ,eresp,"
"wretr,wredis,"
"status,weight,act,bck,"
"chkfail,chkdown,lastchg,downtime,qlimit,"
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
"pid,iid,sid,throttle,lbtot,tracked,type,"
"rate,rate_lim,rate_max,"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
"check_status,check_code,check_duration,"
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
"hrsp_1xx,hrsp_2xx,hrsp_3xx,hrsp_4xx,hrsp_5xx,hrsp_other,hanafail,"
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
"req_rate,req_rate_max,req_tot,"
"cli_abrt,srv_abrt,"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"\n");
}
[MINOR] Break out dumping table This will allow the code to be reused for showing a table filtered by key Signed-off-by: Simon Horman <horms@verge.net.au>
2011-06-15 02:18:45 -04:00
/* print a string of text buffer to <out>. The format is :
* Non-printable chars \t, \n, \r and \e are * encoded in C format.
* Other non-printable chars are encoded "\xHH". Space and '\' are also escaped.
* Print stopped if null char or <bsize> is reached, or if no more place in the chunk.
*/
static int dump_text(struct chunk *out, const char *buf, int bsize)
{
unsigned char c;
int ptr = 0;
while (buf[ptr] && ptr < bsize) {
c = buf[ptr];
if (isprint(c) && isascii(c) && c != '\\' && c != ' ') {
if (out->len > out->size - 1)
break;
out->str[out->len++] = c;
}
else if (c == '\t' || c == '\n' || c == '\r' || c == '\e' || c == '\\' || c == ' ') {
if (out->len > out->size - 2)
break;
out->str[out->len++] = '\\';
switch (c) {
case ' ': c = ' '; break;
case '\t': c = 't'; break;
case '\n': c = 'n'; break;
case '\r': c = 'r'; break;
case '\e': c = 'e'; break;
case '\\': c = '\\'; break;
}
out->str[out->len++] = c;
}
else {
if (out->len > out->size - 4)
break;
out->str[out->len++] = '\\';
out->str[out->len++] = 'x';
out->str[out->len++] = hextab[(c >> 4) & 0xF];
out->str[out->len++] = hextab[c & 0xF];
}
ptr++;
}
return ptr;
}
/* print a buffer in hexa.
* Print stopped if <bsize> is reached, or if no more place in the chunk.
*/
static int dump_binary(struct chunk *out, const char *buf, int bsize)
{
unsigned char c;
int ptr = 0;
while (ptr < bsize) {
c = buf[ptr];
if (out->len > out->size - 2)
break;
out->str[out->len++] = hextab[(c >> 4) & 0xF];
out->str[out->len++] = hextab[c & 0xF];
ptr++;
}
return ptr;
}
/* Dump the status of a table to a stream interface's
* read buffer. It returns 0 if the output buffer is full
* and needs to be called again, otherwise non-zero.
*/
static int stats_dump_table_head_to_buffer(struct chunk *msg, struct stream_interface *si,
struct proxy *proxy, struct proxy *target)
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
[MINOR] Break out dumping table This will allow the code to be reused for showing a table filtered by key Signed-off-by: Simon Horman <horms@verge.net.au>
2011-06-15 02:18:45 -04:00
chunk_printf(msg, "# table: %s, type: %s, size:%d, used:%d\n",
proxy->id, stktable_types[proxy->table.type].kw, proxy->table.size, proxy->table.current);
/* any other information should be dumped here */
if (target && s->listener->perm.ux.level < ACCESS_LVL_OPER)
chunk_printf(msg, "# contents not dumped due to insufficient privileges\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, msg) == -1)
[MINOR] Break out dumping table This will allow the code to be reused for showing a table filtered by key Signed-off-by: Simon Horman <horms@verge.net.au>
2011-06-15 02:18:45 -04:00
return 0;
return 1;
}
/* Dump the a table entry to a stream interface's
* read buffer. It returns 0 if the output buffer is full
* and needs to be called again, otherwise non-zero.
*/
static int stats_dump_table_entry_to_buffer(struct chunk *msg, struct stream_interface *si,
struct proxy *proxy, struct stksess *entry)
{
int dt;
chunk_printf(msg, "%p:", entry);
if (proxy->table.type == STKTABLE_TYPE_IP) {
char addr[INET_ADDRSTRLEN];
inet_ntop(AF_INET, (const void *)&entry->key.key, addr, sizeof(addr));
chunk_printf(msg, " key=%s", addr);
}
else if (proxy->table.type == STKTABLE_TYPE_IPV6) {
char addr[INET6_ADDRSTRLEN];
inet_ntop(AF_INET6, (const void *)&entry->key.key, addr, sizeof(addr));
chunk_printf(msg, " key=%s", addr);
}
else if (proxy->table.type == STKTABLE_TYPE_INTEGER) {
chunk_printf(msg, " key=%u", *(unsigned int *)entry->key.key);
}
else if (proxy->table.type == STKTABLE_TYPE_STRING) {
chunk_printf(msg, " key=");
dump_text(msg, (const char *)entry->key.key, proxy->table.key_size);
}
else {
chunk_printf(msg, " key=");
dump_binary(msg, (const char *)entry->key.key, proxy->table.key_size);
}
chunk_printf(msg, " use=%d exp=%d", entry->ref_cnt - 1, tick_remain(now_ms, entry->expire));
for (dt = 0; dt < STKTABLE_DATA_TYPES; dt++) {
void *ptr;
if (proxy->table.data_ofs[dt] == 0)
continue;
if (stktable_data_types[dt].arg_type == ARG_T_DELAY)
chunk_printf(msg, " %s(%d)=", stktable_data_types[dt].name, proxy->table.data_arg[dt].u);
else
chunk_printf(msg, " %s=", stktable_data_types[dt].name);
ptr = stktable_data_ptr(&proxy->table, entry, dt);
switch (stktable_data_types[dt].std_type) {
case STD_T_SINT:
chunk_printf(msg, "%d", stktable_data_cast(ptr, std_t_sint));
break;
case STD_T_UINT:
chunk_printf(msg, "%u", stktable_data_cast(ptr, std_t_uint));
break;
case STD_T_ULL:
chunk_printf(msg, "%lld", stktable_data_cast(ptr, std_t_ull));
break;
case STD_T_FRQP:
chunk_printf(msg, "%d",
read_freq_ctr_period(&stktable_data_cast(ptr, std_t_frqp),
proxy->table.data_arg[dt].u));
break;
}
}
chunk_printf(msg, "\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, msg) == -1)
[MINOR] Break out dumping table This will allow the code to be reused for showing a table filtered by key Signed-off-by: Simon Horman <horms@verge.net.au>
2011-06-15 02:18:45 -04:00
return 0;
return 1;
}
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
static void stats_sock_table_key_request(struct stream_interface *si, char **args, int action)
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
struct proxy *px = si->applet.ctx.table.target;
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
struct stksess *ts;
[MINOR] Allow showing and clearing by key of integer stick tables
2011-06-15 02:18:51 -04:00
uint32_t uint32_key;
[MINOR] Allow showing and clearing by key of ipv6 stick tables
2011-06-15 02:18:50 -04:00
unsigned char ip6_key[sizeof(struct in6_addr)];
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
struct chunk msg;
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
long long value;
int data_type;
void *ptr;
struct freq_ctr_period *frqp;
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
si->applet.st0 = STAT_CLI_OUTPUT;
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
if (!*args[4]) {
si->applet.ctx.cli.msg = "Key value expected\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
[MINOR] Allow showing and clearing by key of ipv6 stick tables
2011-06-15 02:18:50 -04:00
switch (px->table.type) {
case STKTABLE_TYPE_IP:
[MINOR] Allow showing and clearing by key of integer stick tables
2011-06-15 02:18:51 -04:00
uint32_key = htonl(inetaddr_host(args[4]));
static_table_key.key = &uint32_key;
[MINOR] Allow showing and clearing by key of ipv6 stick tables
2011-06-15 02:18:50 -04:00
break;
case STKTABLE_TYPE_IPV6:
inet_pton(AF_INET6, args[4], ip6_key);
static_table_key.key = &ip6_key;
break;
[MINOR] Allow showing and clearing by key of integer stick tables
2011-06-15 02:18:51 -04:00
case STKTABLE_TYPE_INTEGER:
{
char *endptr;
unsigned long val;
errno = 0;
val = strtoul(args[4], &endptr, 10);
if ((errno == ERANGE && val == ULONG_MAX) ||
(errno != 0 && val == 0) || endptr == args[4] ||
val > 0xffffffff) {
si->applet.ctx.cli.msg = "Invalid key\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
uint32_key = (uint32_t) val;
static_table_key.key = &uint32_key;
break;
}
break;
[MINOR] Allow showing and clearing by key of string stick tables
2011-06-15 02:18:52 -04:00
case STKTABLE_TYPE_STRING:
static_table_key.key = args[4];
static_table_key.key_len = strlen(args[4]);
break;
[MINOR] Allow showing and clearing by key of ipv6 stick tables
2011-06-15 02:18:50 -04:00
default:
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
switch (action) {
case STAT_CLI_O_TAB:
si->applet.ctx.cli.msg = "Showing keys from tables of type other than ip, ipv6, string and integer is not supported\n";
break;
case STAT_CLI_O_CLR:
si->applet.ctx.cli.msg = "Removing keys from ip tables of type other than ip, ipv6, string and integer is not supported\n";
break;
default:
si->applet.ctx.cli.msg = "Unknown action\n";
break;
}
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
si->applet.st0 = STAT_CLI_PRINT;
return;
}
/* check permissions */
if (s->listener->perm.ux.level < ACCESS_LVL_OPER) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return;
}
ts = stktable_lookup_key(&px->table, &static_table_key);
[MINOR] Allow listing of stick table by key
2011-06-15 02:18:47 -04:00
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
switch (action) {
case STAT_CLI_O_TAB:
if (!ts)
return;
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MINOR] Allow listing of stick table by key
2011-06-15 02:18:47 -04:00
if (!stats_dump_table_head_to_buffer(&msg, si, px, px))
return;
stats_dump_table_entry_to_buffer(&msg, si, px, ts);
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
return;
[MINOR] Allow listing of stick table by key
2011-06-15 02:18:47 -04:00
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
case STAT_CLI_O_CLR:
if (!ts)
return;
if (ts->ref_cnt) {
/* don't delete an entry which is currently referenced */
si->applet.ctx.cli.msg = "Entry currently in use, cannot remove\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
stksess_kill(&px->table, ts);
break;
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
case STAT_CLI_O_SET:
if (strncmp(args[5], "data.", 5) != 0) {
si->applet.ctx.cli.msg = "\"data.<type>\" followed by a value expected\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
data_type = stktable_get_data_type(args[5] + 5);
if (data_type < 0) {
si->applet.ctx.cli.msg = "Unknown data type\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
if (!px->table.data_ofs[data_type]) {
si->applet.ctx.cli.msg = "Data type not stored in this table\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
if (!*args[6] || strl2llrc(args[6], strlen(args[6]), &value) != 0) {
si->applet.ctx.cli.msg = "Require a valid integer value to store\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
if (ts)
stktable_touch(&px->table, ts, 1);
else {
ts = stksess_new(&px->table, &static_table_key);
if (!ts) {
/* don't delete an entry which is currently referenced */
si->applet.ctx.cli.msg = "Unable to allocate a new entry\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
stktable_store(&px->table, ts, 1);
}
ptr = stktable_data_ptr(&px->table, ts, data_type);
switch (stktable_data_types[data_type].std_type) {
case STD_T_SINT:
stktable_data_cast(ptr, std_t_sint) = value;
break;
case STD_T_UINT:
stktable_data_cast(ptr, std_t_uint) = value;
break;
case STD_T_ULL:
stktable_data_cast(ptr, std_t_ull) = value;
break;
case STD_T_FRQP:
/* We only reset the previous value so that it slowly fades out */
frqp = &stktable_data_cast(ptr, std_t_frqp);
frqp->curr_tick = now_ms;
frqp->prev_ctr = value;
frqp->curr_ctr = 0;
break;
}
break;
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
default:
si->applet.ctx.cli.msg = "Unknown action\n";
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
si->applet.st0 = STAT_CLI_PRINT;
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
break;
[MINOR] Break out processing of clear table This will allow the code to be reused for showing a table filtered by key
2011-06-15 02:18:46 -04:00
}
}
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
static void stats_sock_table_data_request(struct stream_interface *si, char **args, int action)
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
{
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
if (action != STAT_CLI_O_TAB) {
si->applet.ctx.cli.msg = "content-based lookup is only supported with the \"show\" action";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
/* condition on stored data value */
si->applet.ctx.table.data_type = stktable_get_data_type(args[3] + 5);
if (si->applet.ctx.table.data_type < 0) {
si->applet.ctx.cli.msg = "Unknown data type\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
if (!((struct proxy *)si->applet.ctx.table.target)->table.data_ofs[si->applet.ctx.table.data_type]) {
si->applet.ctx.cli.msg = "Data type not stored in this table\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
si->applet.ctx.table.data_op = get_std_op(args[4]);
if (si->applet.ctx.table.data_op < 0) {
si->applet.ctx.cli.msg = "Require and operator among \"eq\", \"ne\", \"le\", \"ge\", \"lt\", \"gt\"\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
if (!*args[5] || strl2llrc(args[5], strlen(args[5]), &si->applet.ctx.table.value) != 0) {
si->applet.ctx.cli.msg = "Require a valid integer value to compare against\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
}
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
static void stats_sock_table_request(struct stream_interface *si, char **args, int action)
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
{
si->applet.ctx.table.data_type = -1;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INIT;
[BUG] possible crash in 'show table' on stats socket Patch d5b9fd95 was missing an initialisation of "ctx.table.target", which caused "show table" to segfault if it was issued after a "show errors" (target pointer == -1).
2011-08-24 02:23:34 -04:00
si->applet.ctx.table.target = NULL;
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
si->applet.ctx.table.proxy = NULL;
si->applet.ctx.table.entry = NULL;
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
si->applet.st0 = action;
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
if (*args[2]) {
si->applet.ctx.table.target = find_stktable(args[2]);
if (!si->applet.ctx.table.target) {
si->applet.ctx.cli.msg = "No such table\n";
si->applet.st0 = STAT_CLI_PRINT;
return;
}
}
else {
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
if (action != STAT_CLI_O_TAB)
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
goto err_args;
return;
}
if (strcmp(args[3], "key") == 0)
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
stats_sock_table_key_request(si, args, action);
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
else if (strncmp(args[3], "data.", 5) == 0)
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
stats_sock_table_data_request(si, args, action);
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
else if (*args[3])
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
goto err_args;
return;
err_args:
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
switch (action) {
case STAT_CLI_O_TAB:
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
si->applet.ctx.cli.msg = "Optional argument only supports \"data.<store_data_type>\" <operator> <value> and key <key>\n";
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
break;
case STAT_CLI_O_CLR:
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
si->applet.ctx.cli.msg = "Required arguments: <table> \"data.<store_data_type>\" <operator> <value> or <table> key <key>\n";
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
break;
default:
si->applet.ctx.cli.msg = "Unknown action\n";
break;
}
[MINOR] Break out all stick table socat command parsing This will allow reuse for clearing table entries other than by key
2011-06-15 02:18:48 -04:00
si->applet.st0 = STAT_CLI_PRINT;
}
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
/* Expects to find a frontend named <arg> and returns it, otherwise displays various
* adequate error messages and returns NULL. This function also expects the session
* level to be admin.
*/
static struct proxy *expect_frontend_admin(struct session *s, struct stream_interface *si, const char *arg)
{
struct proxy *px;
if (s->listener->perm.ux.level < ACCESS_LVL_ADMIN) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
if (!*arg) {
si->applet.ctx.cli.msg = "A frontend name is expected.\n";
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
px = findproxy(arg, PR_CAP_FE);
if (!px) {
si->applet.ctx.cli.msg = "No such frontend.\n";
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
return px;
}
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
/* Expects to find a backend and a server in <arg> under the form <backend>/<server>,
* and returns the pointer to the server. Otherwise, display adequate error messages
* and returns NULL. This function also expects the session level to be admin. Note:
* the <arg> is modified to remove the '/'.
*/
static struct server *expect_server_admin(struct session *s, struct stream_interface *si, char *arg)
{
struct proxy *px;
struct server *sv;
char *line;
if (s->listener->perm.ux.level < ACCESS_LVL_ADMIN) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
/* split "backend/server" and make <line> point to server */
for (line = arg; *line; line++)
if (*line == '/') {
*line++ = '\0';
break;
}
if (!*line || !*arg) {
si->applet.ctx.cli.msg = "Require 'backend/server'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
if (!get_backend_server(arg, line, &px, &sv)) {
si->applet.ctx.cli.msg = px ? "No such server.\n" : "No such backend.\n";
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
if (px->state == PR_STSTOPPED) {
si->applet.ctx.cli.msg = "Proxy is disabled.\n";
si->applet.st0 = STAT_CLI_PRINT;
return NULL;
}
return sv;
}
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* Processes the stats interpreter on the statistics socket. This function is
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
* called from an applet running in a stream interface. The function returns 1
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
* if the request was understood, otherwise zero. It sets si->applet.st0 to a value
[MINOR] stats: use a dedicated state to output static data It is a bit expensive and complex to use to call buffer_feed() directly from the request parser, and there are risks that some output messages are lost in case of buffer full. Since most of these messages are static, let's have a state dedicated to print these messages and store them in a specific area shared with the stats in the session. This both reduces code size and risks of losing output data.
2009-10-11 17:12:51 -04:00
* designating the function which will have to process the request, which can
* also be the print function to display the return message set into cli.msg.
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_sock_parse_request(struct stream_interface *si, char *line)
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
char *args[MAX_STATS_ARGS + 1];
int arg;
while (isspace((unsigned char)*line))
line++;
arg = 0;
args[arg] = line;
while (*line && arg < MAX_STATS_ARGS) {
if (isspace((unsigned char)*line)) {
*line++ = '\0';
while (isspace((unsigned char)*line))
line++;
args[++arg] = line;
continue;
}
line++;
}
while (++arg <= MAX_STATS_ARGS)
args[arg] = line;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.flags = 0;
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
if (strcmp(args[0], "show") == 0) {
if (strcmp(args[1], "stat") == 0) {
if (*args[2] && *args[3] && *args[4]) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.flags |= STAT_BOUND;
si->applet.ctx.stats.iid = atoi(args[2]);
si->applet.ctx.stats.type = atoi(args[3]);
si->applet.ctx.stats.sid = atoi(args[4]);
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.flags |= STAT_SHOW_STAT;
si->applet.ctx.stats.flags |= STAT_FMT_CSV;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INIT;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_O_INFO; // stats_dump_raw_to_buffer
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
else if (strcmp(args[1], "info") == 0) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.flags |= STAT_SHOW_INFO;
si->applet.ctx.stats.flags |= STAT_FMT_CSV;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INIT;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_O_INFO; // stats_dump_raw_to_buffer
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
else if (strcmp(args[1], "sess") == 0) {
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INIT;
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
if (s->listener->perm.ux.level < ACCESS_LVL_OPER) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = stats_permission_denied_msg;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
return 1;
}
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
if (*args[2])
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.target = (void *)strtoul(args[2], NULL, 0);
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
else
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.target = NULL;
si->applet.ctx.sess.section = 0; /* start with session status */
si->applet.ctx.sess.pos = 0;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_O_SESS; // stats_dump_sess_to_buffer
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
else if (strcmp(args[1], "errors") == 0) {
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
if (s->listener->perm.ux.level < ACCESS_LVL_OPER) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = stats_permission_denied_msg;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
return 1;
}
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
if (*args[2])
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.iid = atoi(args[2]);
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
else
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.iid = -1;
si->applet.ctx.errors.px = NULL;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INIT;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_O_ERR; // stats_dump_errors_to_buffer
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
else if (strcmp(args[1], "table") == 0) {
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
stats_sock_table_request(si, args, STAT_CLI_O_TAB);
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
}
CLEANUP: Fix some minor typos
2012-04-02 21:57:54 -04:00
else { /* neither "stat" nor "info" nor "sess" nor "errors" nor "table" */
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
return 0;
}
}
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
else if (strcmp(args[0], "clear") == 0) {
if (strcmp(args[1], "counters") == 0) {
struct proxy *px;
struct server *sv;
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
struct listener *li;
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
int clrall = 0;
if (strcmp(args[2], "all") == 0)
clrall = 1;
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
/* check permissions */
if (s->listener->perm.ux.level < ACCESS_LVL_OPER ||
(clrall && s->listener->perm.ux.level < ACCESS_LVL_ADMIN)) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = stats_permission_denied_msg;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] add access restrictions to the stats socket The stats socket can now run at 3 different levels : - user - operator (default one) - admin These levels are used to restrict access to some information and commands. Only the admin can clear all stats. A user cannot clear anything nor access sensible data such as sessions or errors.
2009-10-10 11:13:00 -04:00
return 1;
}
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
for (px = proxy; px; px = px->next) {
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
if (clrall) {
memset(&px->be_counters, 0, sizeof(px->be_counters));
memset(&px->fe_counters, 0, sizeof(px->fe_counters));
}
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
else {
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->be_counters.conn_max = 0;
px->be_counters.p.http.rps_max = 0;
px->be_counters.sps_max = 0;
px->be_counters.cps_max = 0;
px->be_counters.nbpend_max = 0;
px->fe_counters.conn_max = 0;
px->fe_counters.p.http.rps_max = 0;
px->fe_counters.sps_max = 0;
px->fe_counters.cps_max = 0;
px->fe_counters.nbpend_max = 0;
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
}
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
for (sv = px->srv; sv; sv = sv->next)
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
if (clrall)
memset(&sv->counters, 0, sizeof(sv->counters));
else {
sv->counters.cur_sess_max = 0;
sv->counters.nbpend_max = 0;
sv->counters.sps_max = 0;
}
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
for (li = px->listen; li; li = li->next)
[MINOR] stats: use 'clear counters all' to clear all values The most common use of "clear counters" should be to only clear max values without affecting cumulated values, for instance, after an incident. So we change "clear counters" to only clear max values, and add "clear counters all" to clear all counters.
2009-10-10 09:26:26 -04:00
if (li->counters) {
if (clrall)
memset(li->counters, 0, sizeof(*li->counters));
else
li->counters->conn_max = 0;
}
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
}
[MEDIUM] add support for global.maxconnrate to limit the per-process conn rate. This one enforces a per-process connection rate limit, regardless of what may be set per frontend. It can be a way to limit the CPU usage of a process being severely attacked. The side effect is that the global process connection rate is now measured for each incoming connection, so it will be possible to report it.
2011-09-07 09:17:21 -04:00
global.cps_max = 0;
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
return 1;
}
[MEDIUM] stats: add "clear table <name> key <value>" to clear table entries This feature will be required at some point, when the stick tables are used to enforce security measures. For instance, some visitors may be incorrectly flagged as abusers and would ask the site admins to remove their entry from the table.
2010-07-13 07:48:00 -04:00
else if (strcmp(args[1], "table") == 0) {
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
stats_sock_table_request(si, args, STAT_CLI_O_CLR);
[MEDIUM] stats: add "clear table <name> key <value>" to clear table entries This feature will be required at some point, when the stick tables are used to enforce security measures. For instance, some visitors may be incorrectly flagged as abusers and would ask the site admins to remove their entry from the table.
2010-07-13 07:48:00 -04:00
/* end of processing */
return 1;
}
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
else {
[MEDIUM] stats: add "clear table <name> key <value>" to clear table entries This feature will be required at some point, when the stick tables are used to enforce security measures. For instance, some visitors may be incorrectly flagged as abusers and would ask the site admins to remove their entry from the table.
2010-07-13 07:48:00 -04:00
/* unknown "clear" argument */
[MINOR] Add "clear counters" to clear statistics counters Now, when statistics counters are moved to separate structures, adding "clear counters" is extremely easy. [wt: updated the help message]
2009-10-04 09:02:46 -04:00
return 0;
}
}
[MEDIUM] stats: add the "get weight" command to report a server's weight This command is unrestricted and simply reports a server's weight, as well as the weight initially configured, which at the moment are the sames.
2009-10-10 12:37:29 -04:00
else if (strcmp(args[0], "get") == 0) {
if (strcmp(args[1], "weight") == 0) {
struct proxy *px;
struct server *sv;
/* split "backend/server" and make <line> point to server */
for (line = args[2]; *line; line++)
if (*line == '/') {
*line++ = '\0';
break;
}
if (!*line) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Require 'backend/server'.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] stats: add the "get weight" command to report a server's weight This command is unrestricted and simply reports a server's weight, as well as the weight initially configured, which at the moment are the sames.
2009-10-10 12:37:29 -04:00
return 1;
}
if (!get_backend_server(args[2], line, &px, &sv)) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = px ? "No such server.\n" : "No such backend.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] stats: add the "get weight" command to report a server's weight This command is unrestricted and simply reports a server's weight, as well as the weight initially configured, which at the moment are the sames.
2009-10-10 12:37:29 -04:00
return 1;
}
/* return server's effective weight at the moment */
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
snprintf(trash, trashlen, "%d (initial %d)\n", sv->uweight, sv->iweight);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
bi_putstr(si->ib, trash);
[MEDIUM] stats: add the "get weight" command to report a server's weight This command is unrestricted and simply reports a server's weight, as well as the weight initially configured, which at the moment are the sames.
2009-10-10 12:37:29 -04:00
return 1;
}
else { /* not "get weight" */
return 0;
}
}
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
else if (strcmp(args[0], "set") == 0) {
if (strcmp(args[1], "weight") == 0) {
struct proxy *px;
struct server *sv;
int w;
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
sv = expect_server_admin(s, si, args[2]);
if (!sv)
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 1;
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
px = sv->proxy;
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
/* if the weight is terminated with '%', it is set relative to
* the initial weight, otherwise it is absolute.
*/
if (!*args[3]) {
si->applet.ctx.cli.msg = "Require <weight> or <weight%>.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[BUG] stats: admin commands must check the proxy state As reported by Bryan Talbot, enabling and disabling a server in a disabled proxy causes a segfault. Changing the weight can also cause a similar segfault.
2011-03-03 14:49:04 -05:00
return 1;
}
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
w = atoi(args[3]);
if (strchr(args[3], '%') != NULL) {
if (w < 0 || w > 100) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Relative weight can only be set between 0 and 100% inclusive.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 1;
}
w = sv->iweight * w / 100;
}
else {
if (w < 0 || w > 256) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Absolute weight can only be between 0 and 256 inclusive.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 1;
}
}
if (w && w != sv->iweight && !(px->lbprm.algo & BE_LB_PROP_DYN)) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Backend is using a static LB algorithm and only accepts weights '0%' and '100%'.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 1;
}
sv->uweight = w;
if (px->lbprm.algo & BE_LB_PROP_DYN) {
/* we must take care of not pushing the server to full throttle during slow starts */
if ((sv->state & SRV_WARMINGUP) && (px->lbprm.algo & BE_LB_PROP_DYN))
sv->eweight = (BE_WEIGHT_SCALE * (now.tv_sec - sv->last_change) + sv->slowstart - 1) / sv->slowstart;
else
sv->eweight = BE_WEIGHT_SCALE;
sv->eweight *= sv->uweight;
} else {
sv->eweight = sv->uweight;
}
/* static LB algorithms are a bit harder to update */
if (px->lbprm.update_server_eweight)
px->lbprm.update_server_eweight(sv);
BUG/MAJOR: checks: don't call set_server_status_* when no LB algo is set David Touzeau reported that haproxy dies when a server is checked and is used in a farm with only "option transparent" and no LB algo. This is because the LB params are NULL, the functions should be checked before being called. The same bug is present in 1.4 so this patch must be backported.
2012-05-19 13:07:40 -04:00
else if (sv->eweight) {
if (px->lbprm.set_server_status_up)
px->lbprm.set_server_status_up(sv);
}
else {
if (px->lbprm.set_server_status_down)
px->lbprm.set_server_status_down(sv);
}
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 1;
}
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
else if (strcmp(args[1], "timeout") == 0) {
if (strcmp(args[2], "cli") == 0) {
unsigned timeout;
const char *res;
if (!*args[3]) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Expects an integer value.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
return 1;
}
res = parse_time_err(args[3], &timeout, TIME_UNIT_S);
if (res || timeout < 1) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "Invalid timeout value.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
return 1;
}
s->req->rto = s->rep->wto = 1 + MS_TO_TICKS(timeout*1000);
return 1;
}
else {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = "'set timeout' only supports 'cli'.\n";
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
return 1;
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
}
}
else if (strcmp(args[1], "maxconn") == 0) {
if (strcmp(args[2], "frontend") == 0) {
struct proxy *px;
struct listener *l;
int v;
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
px = expect_frontend_admin(s, si, args[3]);
if (!px)
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
return 1;
if (!*args[4]) {
si->applet.ctx.cli.msg = "Integer value expected.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
v = atoi(args[4]);
/* check for unlimited values, we restore default setting (cfg_maxpconn) */
if (v < 1) {
si->applet.ctx.cli.msg = "Value out of range.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
/* OK, the value is fine, so we assign it to the proxy and to all of
* its listeners. The blocked ones will be dequeued.
*/
px->maxconn = v;
for (l = px->listen; l != NULL; l = l->next) {
l->maxconn = v;
if (l->state == LI_FULL)
resume_listener(l);
}
if (px->maxconn > px->feconn && !LIST_ISEMPTY(&s->fe->listener_queue))
dequeue_all_listeners(&s->fe->listener_queue);
return 1;
}
[MEDIUM] stats: add the "set maxconn" setting to the command line interface This option permits to change the global maxconn setting within the limit that was set by the initial value, which is now reported as the hard maxconn value. This allows to immediately accept more concurrent connections or to stop accepting new ones until the value passes below the indicated setting. The main use of this option is on systems where many haproxy instances are loaded and admins need to re-adjust resource sharing at run time to regain a bit of fairness between processes.
2011-09-07 08:38:31 -04:00
else if (strcmp(args[2], "global") == 0) {
int v;
if (s->listener->perm.ux.level < ACCESS_LVL_ADMIN) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (!*args[3]) {
si->applet.ctx.cli.msg = "Expects an integer value.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
v = atoi(args[3]);
if (v > global.hardmaxconn) {
si->applet.ctx.cli.msg = "Value out of range.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
/* check for unlimited values */
if (v <= 0)
v = global.hardmaxconn;
global.maxconn = v;
/* Dequeues all of the listeners waiting for a resource */
if (!LIST_ISEMPTY(&global_listener_queue))
dequeue_all_listeners(&global_listener_queue);
return 1;
}
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
else {
[MEDIUM] stats: add the "set maxconn" setting to the command line interface This option permits to change the global maxconn setting within the limit that was set by the initial value, which is now reported as the hard maxconn value. This allows to immediately accept more concurrent connections or to stop accepting new ones until the value passes below the indicated setting. The main use of this option is on systems where many haproxy instances are loaded and admins need to re-adjust resource sharing at run time to regain a bit of fairness between processes.
2011-09-07 08:38:31 -04:00
si->applet.ctx.cli.msg = "'set maxconn' only supports 'frontend' and 'global'.\n";
[MEDIUM] stats: add support for changing frontend's maxconn at runtime The new "set maxconn frontend XXX" statement on the stats socket allows the admin to change a frontend's maxconn value. If some connections are queued, they will immediately be accepted up to the new limit. If the limit is lowered, new connections acceptation might be delayed. This can be used to temporarily reduce or increase the impact of a specific frontend's traffic on the whole process.
2011-08-02 05:49:05 -04:00
si->applet.st0 = STAT_CLI_PRINT;
return 1;
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
}
}
[MEDIUM] stats: add the ability to adjust the global maxconnrate Using "set rate-limit connections global <xxx>" on the CLI, we can now adjust the per-process connection rate limiting (equal to global.maxconnrate).
2011-09-07 10:13:44 -04:00
else if (strcmp(args[1], "rate-limit") == 0) {
if (strcmp(args[2], "connections") == 0) {
if (strcmp(args[3], "global") == 0) {
int v;
if (s->listener->perm.ux.level < ACCESS_LVL_ADMIN) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (!*args[4]) {
si->applet.ctx.cli.msg = "Expects an integer value.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
v = atoi(args[4]);
if (v < 0) {
si->applet.ctx.cli.msg = "Value out of range.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
global.cps_lim = v;
/* Dequeues all of the listeners waiting for a resource */
if (!LIST_ISEMPTY(&global_listener_queue))
dequeue_all_listeners(&global_listener_queue);
return 1;
}
else {
si->applet.ctx.cli.msg = "'set rate-limit connections' only supports 'global'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
}
else {
si->applet.ctx.cli.msg = "'set rate-limit' only supports 'connections'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
}
MEDIUM: stats/cli: add support for "set table key" to enter values This is used to enter values for stick tables. The most likely usage is to set gpc0 for a specific IP address in order to block traffic for abusers without having to reload. Since all data types are supported, other usages are possible (eg: replace a users's assigned server).
2012-06-06 19:03:16 -04:00
else if (strcmp(args[1], "table") == 0) {
stats_sock_table_request(si, args, STAT_CLI_O_SET);
}
[STATS] make it possible to change a CLI connection timeout Sometimes it helps to be able to change an interactive CLI connection timeout. Now we just have to enter "set timeout cli <value>" to do that.
2010-01-26 04:59:06 -05:00
else { /* unknown "set" parameter */
[MEDIUM] stats: add the "set weight" command It is now possible to change a server's weight from the stats socket. Just use "set weight <back>/<serv> <weight>".
2009-10-10 13:30:08 -04:00
return 0;
}
}
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
else if (strcmp(args[0], "enable") == 0) {
if (strcmp(args[1], "server") == 0) {
struct server *sv;
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
sv = expect_server_admin(s, si, args[2]);
if (!sv)
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
return 1;
[BUG] stats: admin commands must check the proxy state As reported by Bryan Talbot, enabling and disabling a server in a disabled proxy causes a segfault. Changing the weight can also cause a similar segfault.
2011-03-03 14:49:04 -05:00
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
if (sv->state & SRV_MAINTAIN) {
/* The server is really in maintenance, we can change the server state */
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
if (sv->track) {
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
/* If this server tracks the status of another one,
* we must restore the good status.
*/
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
if (sv->track->state & SRV_RUNNING) {
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
set_server_up(sv);
[MEDIUM] checks: set server state to one state from failure when leaving maintenance When we're enabling a server again (unix CLI or stats interface), we must not mark it completely up because it can take a while before a failure is detected. So we mark it one step above failure, which means it's up but will be marked down upon first failure. (cherry picked from commit 83c3e06452457ed5660fc814cbda5bf878bf19a2)
2010-10-22 08:39:02 -04:00
sv->health = sv->rise; /* up, but will fall down at first failure */
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
} else {
sv->state &= ~SRV_MAINTAIN;
set_server_down(sv);
}
} else {
set_server_up(sv);
[MEDIUM] checks: set server state to one state from failure when leaving maintenance When we're enabling a server again (unix CLI or stats interface), we must not mark it completely up because it can take a while before a failure is detected. So we mark it one step above failure, which means it's up but will be marked down upon first failure. (cherry picked from commit 83c3e06452457ed5660fc814cbda5bf878bf19a2)
2010-10-22 08:39:02 -04:00
sv->health = sv->rise; /* up, but will fall down at first failure */
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
}
}
return 1;
}
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
else if (strcmp(args[1], "frontend") == 0) {
struct proxy *px;
px = expect_frontend_admin(s, si, args[2]);
if (!px)
return 1;
if (px->state == PR_STSTOPPED) {
si->applet.ctx.cli.msg = "Frontend was previously shut down, cannot enable.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (px->state != PR_STPAUSED) {
si->applet.ctx.cli.msg = "Frontend is already enabled.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (!resume_proxy(px)) {
si->applet.ctx.cli.msg = "Failed to resume frontend, check logs for precise cause (port conflict?).\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
return 1;
}
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
else { /* unknown "enable" parameter */
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
si->applet.ctx.cli.msg = "'enable' only supports 'frontend' and 'server'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
}
}
else if (strcmp(args[0], "disable") == 0) {
if (strcmp(args[1], "server") == 0) {
struct server *sv;
[CLEANUP] stats: centralize tests for backend/server inputs on the CLI The tests were repeated many times. Let's put them at one single place.
2011-09-07 17:41:01 -04:00
sv = expect_server_admin(s, si, args[2]);
if (!sv)
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
return 1;
[BUG] stats: admin commands must check the proxy state As reported by Bryan Talbot, enabling and disabling a server in a disabled proxy causes a segfault. Changing the weight can also cause a similar segfault.
2011-03-03 14:49:04 -05:00
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
if (! (sv->state & SRV_MAINTAIN)) {
/* Not already in maintenance, we can change the server state */
sv->state |= SRV_MAINTAIN;
set_server_down(sv);
}
return 1;
}
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
else if (strcmp(args[1], "frontend") == 0) {
struct proxy *px;
px = expect_frontend_admin(s, si, args[2]);
if (!px)
return 1;
if (px->state == PR_STSTOPPED) {
si->applet.ctx.cli.msg = "Frontend was previously shut down, cannot disable.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (px->state == PR_STPAUSED) {
si->applet.ctx.cli.msg = "Frontend is already disabled.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (!pause_proxy(px)) {
si->applet.ctx.cli.msg = "Failed to pause frontend, check logs for precise cause.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
return 1;
}
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
else { /* unknown "disable" parameter */
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
si->applet.ctx.cli.msg = "'disable' only supports 'frontend' and 'server'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
}
else if (strcmp(args[0], "shutdown") == 0) {
if (strcmp(args[1], "frontend") == 0) {
struct proxy *px;
px = expect_frontend_admin(s, si, args[2]);
if (!px)
return 1;
if (px->state == PR_STSTOPPED) {
si->applet.ctx.cli.msg = "Frontend was already shut down.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
Warning("Proxy %s stopped (FE: %lld conns, BE: %lld conns).\n",
px->id, px->fe_counters.cum_conn, px->be_counters.cum_conn);
send_log(px, LOG_WARNING, "Proxy %s stopped (FE: %lld conns, BE: %lld conns).\n",
px->id, px->fe_counters.cum_conn, px->be_counters.cum_conn);
stop_proxy(px);
return 1;
}
[MEDIUM] stats: offer the possibility to kill a session from the CLI It's now possible to issue "shutdown session 0xXXXXXXXX" and have this session immediately killed. Useful for long-running fantoms.
2011-09-07 17:21:03 -04:00
else if (strcmp(args[1], "session") == 0) {
struct session *sess, *ptr;
if (s->listener->perm.ux.level < ACCESS_LVL_ADMIN) {
si->applet.ctx.cli.msg = stats_permission_denied_msg;
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
if (!*args[2]) {
si->applet.ctx.cli.msg = "Session pointer expected (use 'show sess').\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
ptr = (void *)strtoul(args[2], NULL, 0);
/* first, look for the requested session in the session table */
list_for_each_entry(sess, &sessions, list) {
if (sess == ptr)
break;
}
/* do we have the session ? */
if (sess != ptr) {
si->applet.ctx.cli.msg = "No such session (use 'show sess').\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
session_shutdown(sess, SN_ERR_KILLED);
return 1;
}
[MEDIUM] stats: offer the possibility to kill sessions by server It's now possible to issue "shutdown sessions server <back/srv>" and have all this server's sessions immediately killed.
2011-09-07 17:48:48 -04:00
else if (strcmp(args[1], "sessions") == 0) {
if (strcmp(args[2], "server") == 0) {
struct server *sv;
struct session *sess, *sess_bck;
sv = expect_server_admin(s, si, args[3]);
if (!sv)
return 1;
/* kill all the session that are on this server */
list_for_each_entry_safe(sess, sess_bck, &sv->actconns, by_srv)
if (sess->srv_conn == sv)
session_shutdown(sess, SN_ERR_KILLED);
return 1;
}
else {
si->applet.ctx.cli.msg = "'shutdown sessions' only supports 'server'.\n";
si->applet.st0 = STAT_CLI_PRINT;
return 1;
}
}
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
else { /* unknown "disable" parameter */
[MEDIUM] stats: offer the possibility to kill sessions by server It's now possible to issue "shutdown sessions server <back/srv>" and have all this server's sessions immediately killed.
2011-09-07 17:48:48 -04:00
si->applet.ctx.cli.msg = "'shutdown' only supports 'frontend', 'session' and 'sessions'.\n";
[MEDIUM] stats: add the ability to enable/disable/shutdown a frontend at runtime The stats socket now allows the admin to disable, enable or shutdown a frontend. This can be used when a bug is discovered in a configuration and it's desirable to fix it but the rules in place don't allow to change a running config. Thus it becomes possible to kill the frontend to release the port and start a new one in a separate process. This can also be used to temporarily make haproxy return TCP resets to incoming requests to pretend the service is not bound. For instance, this may be useful to quickly flush a very deep SYN backlog. The frontend check and lookup code was factored with the "set maxconn" usage.
2011-09-07 16:37:44 -04:00
si->applet.st0 = STAT_CLI_PRINT;
return 1;
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
}
}
else { /* not "show" nor "clear" nor "get" nor "set" nor "enable" nor "disable" */
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
return 0;
}
return 1;
}
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* This I/O handler runs as an applet embedded in a stream interface. It is
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
* used to processes I/O from/to the stats unix socket. The system relies on a
* state machine handling requests and various responses. We read a request,
* then we process it and send the response, and we possibly display a prompt.
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
* Then we can read again. The state is stored in si->applet.st0 and is one of the
* STAT_CLI_* constants. si->applet.st1 is used to indicate whether prompt is enabled
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
* or not.
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
*/
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
static void cli_io_handler(struct stream_interface *si)
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
{
REORG/MAJOR: use "struct channel" instead of "struct buffer" This is a massive rename. We'll then split channel and buffer. This change needs a lot of cleanups. At many locations, the parameter or variable is still called "buf" which will become ambiguous. Also, the "struct channel" is still defined in buffers.h.
2012-07-02 09:11:27 -04:00
struct channel *req = si->ob;
struct channel *res = si->ib;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
int reql;
int len;
if (unlikely(si->state == SI_ST_DIS || si->state == SI_ST_CLO))
goto out;
while (1) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if (si->applet.st0 == STAT_CLI_INIT) {
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* Stats output not initialized yet */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
memset(&si->applet.ctx.stats, 0, sizeof(si->applet.ctx.stats));
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_GETREQ;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
}
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
else if (si->applet.st0 == STAT_CLI_END) {
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
/* Let's close for real now. We just close the request
* side, the conditions below will complete if needed.
*/
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutw(si);
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
break;
}
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
else if (si->applet.st0 == STAT_CLI_GETREQ) {
[OPTIM] stats: check free space before trying to print This alone makes a typical HTML stats dump consume 10% CPU less, because we avoid doing complex printf calls to drop them later. Only a few common cases have been checked, those which are very likely to run for nothing.
2009-10-11 17:35:10 -04:00
/* ensure we have some output room left in the event we
* would want to return some info right after parsing.
*/
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
if (buffer_almost_full(&si->ib->buf))
[OPTIM] stats: check free space before trying to print This alone makes a typical HTML stats dump consume 10% CPU less, because we avoid doing complex printf calls to drop them later. Only a few common cases have been checked, those which are very likely to run for nothing.
2009-10-11 17:35:10 -04:00
break;
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
reql = bo_getline(si->ob, trash, trashlen);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
if (reql <= 0) { /* closed or EOL not found */
if (reql == 0)
break;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_END;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
continue;
}
/* seek for a possible semi-colon. If we find one, we
* replace it with an LF and skip only this part.
*/
for (len = 0; len < reql; len++)
if (trash[len] == ';') {
trash[len] = '\n';
reql = len + 1;
break;
}
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
[MINOR] stats: strip CR and LF from the input command line commands sent from certain tools (eg: telnet) will be terminated by CRLF and not only LF. So we must suppress CR too.
2009-10-04 01:36:58 -04:00
/* now it is time to check that we have a full line,
* remove the trailing \n and possibly \r, then cut the
* line.
*/
len = reql - 1;
if (trash[len] != '\n') {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_END;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
continue;
}
[MINOR] stats: strip CR and LF from the input command line commands sent from certain tools (eg: telnet) will be terminated by CRLF and not only LF. So we must suppress CR too.
2009-10-04 01:36:58 -04:00
if (len && trash[len-1] == '\r')
len--;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
trash[len] = '\0';
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
if (len) {
if (strcmp(trash, "quit") == 0) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_END;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
continue;
}
else if (strcmp(trash, "prompt") == 0)
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st1 = !si->applet.st1;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
else if (strcmp(trash, "help") == 0 ||
[MINOR] stats: use a dedicated state to output static data It is a bit expensive and complex to use to call buffer_feed() directly from the request parser, and there are risks that some output messages are lost in case of buffer full. Since most of these messages are static, let's have a state dedicated to print these messages and store them in a specific area shared with the stats in the session. This both reduces code size and risks of losing output data.
2009-10-11 17:12:51 -04:00
!stats_sock_parse_request(si, trash)) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = stats_sock_usage_msg;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[MINOR] stats: use a dedicated state to output static data It is a bit expensive and complex to use to call buffer_feed() directly from the request parser, and there are risks that some output messages are lost in case of buffer full. Since most of these messages are static, let's have a state dedicated to print these messages and store them in a specific area shared with the stats in the session. This both reduces code size and risks of losing output data.
2009-10-11 17:12:51 -04:00
}
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
/* NB: stats_sock_parse_request() may have put
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
* another STAT_CLI_O_* into si->applet.st0.
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
*/
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
}
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
else if (!si->applet.st1) {
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* if prompt is disabled, print help on empty lines,
* so that the user at least knows how to enable
* prompt and find help.
*/
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.cli.msg = stats_sock_usage_msg;
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PRINT;
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* re-adjust req buffer */
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
bo_skip(si->ob, reql);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
req->flags |= BF_READ_DONTWAIT; /* we plan to read small requests */
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
else { /* output functions: first check if the output buffer is closed then abort */
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
if (res->flags & (BF_SHUTR_NOW|BF_SHUTR)) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_END;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
continue;
}
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
switch (si->applet.st0) {
[MINOR] stats: use a dedicated state to output static data It is a bit expensive and complex to use to call buffer_feed() directly from the request parser, and there are risks that some output messages are lost in case of buffer full. Since most of these messages are static, let's have a state dedicated to print these messages and store them in a specific area shared with the stats in the session. This both reduces code size and risks of losing output data.
2009-10-11 17:12:51 -04:00
case STAT_CLI_PRINT:
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putstr(si->ib, si->applet.ctx.cli.msg) != -1)
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
case STAT_CLI_O_INFO:
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_raw_to_buffer(si))
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
case STAT_CLI_O_SESS:
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_sess_to_buffer(si))
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
case STAT_CLI_O_ERR: /* errors dump */
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_errors_to_buffer(si))
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
case STAT_CLI_O_TAB:
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
case STAT_CLI_O_CLR:
MINOR: stats/cli: add plans to support more stick-table actions Right now we only support show/clear on a table. In order to introduce the "set" keyword we need to get rid of the "show" boolean arg. There is no functional change up to this commit.
2012-06-06 17:37:08 -04:00
if (stats_table_request(si, si->applet.st0))
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
default: /* abnormal state */
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_PROMPT;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
}
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
/* The post-command prompt is either LF alone or LF + '> ' in interactive mode */
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if (si->applet.st0 == STAT_CLI_PROMPT) {
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putstr(si->ib, si->applet.st1 ? "\n> " : "\n") != -1)
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_GETREQ;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
}
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
/* If the output functions are still there, it means they require more room. */
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if (si->applet.st0 >= STAT_CLI_OUTPUT)
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
break;
/* Now we close the output if one of the writers did so,
* or if we're not in interactive mode and the request
* buffer is empty. This still allows pipelined requests
* to be sent in non-interactive mode.
*/
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
if ((res->flags & (BF_SHUTW|BF_SHUTW_NOW)) || (!si->applet.st1 && !req->buf.o)) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_END;
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
continue;
}
[MEDIUM] stats: don't use s->ana_state anymore The stats handler used to store internal states in s->ana_state. Now we only rely on si->st0 in which we can store as many states as we have possible outputs. This cleans up the stats code a lot and makes it more maintainable. It has also reduced code size by a few hundred bytes.
2009-10-04 08:22:18 -04:00
/* switch state back to GETREQ to read next requests */
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = STAT_CLI_GETREQ;
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
}
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if ((res->flags & BF_SHUTR) && (si->state == SI_ST_EST) && (si->applet.st0 != STAT_CLI_GETREQ)) {
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
DPRINTF(stderr, "%s@%d: si to buf closed. req=%08x, res=%08x, st=%d\n",
__FUNCTION__, __LINE__, req->flags, res->flags, si->state);
CLEANUP: Fix some minor typos
2012-04-02 21:57:54 -04:00
/* Other side has closed, let's abort if we have no more processing to do
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
* and nothing more to consume. This is comparable to a broken pipe, so
* we forward the close to the request side so that it flows upstream to
* the client.
*/
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutw(si);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
}
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if ((req->flags & BF_SHUTW) && (si->state == SI_ST_EST) && (si->applet.st0 < STAT_CLI_OUTPUT)) {
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
DPRINTF(stderr, "%s@%d: buf to si closed. req=%08x, res=%08x, st=%d\n",
__FUNCTION__, __LINE__, req->flags, res->flags, si->state);
/* We have no more processing to do, and nothing more to send, and
* the client side has closed. So we'll forward this state downstream
* on the response buffer.
*/
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutr(si);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
res->flags |= BF_READ_NULL;
}
/* update all other flags and resync with the other side */
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_update(si);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
/* we don't want to expire timeouts while we're processing requests */
si->ib->rex = TICK_ETERNITY;
si->ob->wex = TICK_ETERNITY;
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
out:
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
DPRINTF(stderr, "%s@%d: st=%d, rqf=%x, rpf=%x, rqh=%d, rqs=%d, rh=%d, rs=%d\n",
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
__FUNCTION__, __LINE__,
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
si->state, req->flags, res->flags, req->i, req->o, res->i, res->o);
[MEDIUM] stats: replace the stats socket analyser with an SI applet We can get rid of the stats analyser by moving all the stats code to a stream interface applet. Above being cleaner, it provides new advantages such as the ability to process requests and responses from the same function and work only with simple state machines. There's no need for any hijack hack anymore. The direct advantage for the user are the interactive mode and the ability to chain several commands delimited by a semi-colon. Now if the user types "prompt", he gets a prompt from which he can send as many requests as he wants. All outputs are terminated by a blank line followed by a new prompt, so this can be used from external tools too. The code is not very clean, it needs some rework, but some part of the dirty parts are due to the remnants of the hijack mode used in the old functions we call. The old AN_REQ_STATS_SOCK analyser flag is now unused and has been removed.
2009-09-22 13:31:03 -04:00
if (unlikely(si->state == SI_ST_DIS || si->state == SI_ST_CLO)) {
/* check that we have released everything then unregister */
stream_int_unregister_handler(si);
[CLEANUP] move remaining stats sockets code to dumpstats The remains of the stats socket code has nothing to do in proto_uxst anymore and must move to dumpstats. The code is much cleaner and more structured. It was also an opportunity to rename AN_REQ_UNIX_STATS as AN_REQ_STATS_SOCK as the stats socket is no longer unix-specific either. The last item refering to stats in proto_uxst is the setting of the task's nice value which should in fact come from the listener.
2009-08-16 13:06:42 -04:00
}
}
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps statistics onto the stream interface's read buffer.
* The data_ctx must have been zeroed first, and the flags properly set.
[MINOR] stats: make stats_dump_raw_to_buffer() use buffer_feed_chunk Same as previous change. A remaining call to stats_dump_proxy() still prevents us from completing the update.
2009-10-04 05:54:04 -04:00
* It returns 0 as long as it does not complete, non-zero upon completion.
* Some states are not used but it makes the code more similar to other
* functions which handle stats too.
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_raw_to_buffer(struct stream_interface *si)
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
{
struct proxy *px;
struct chunk msg;
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
unsigned int up;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
switch (si->conn.data_st) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INIT:
[MINOR] stats: make stats_dump_raw_to_buffer() use buffer_feed_chunk Same as previous change. A remaining call to stats_dump_proxy() still prevents us from completing the update.
2009-10-04 05:54:04 -04:00
/* the function had not been called yet */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_HEAD;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_HEAD:
if (si->applet.ctx.stats.flags & STAT_SHOW_STAT) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
print_csv_header(&msg);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
return 0;
}
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INFO;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INFO:
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
up = (now.tv_sec - start_date.tv_sec);
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_SHOW_INFO) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
"Name: " PRODUCT_NAME "\n"
"Version: " HAPROXY_VERSION "\n"
"Release_date: " HAPROXY_DATE "\n"
"Nbproc: %d\n"
"Process_num: %d\n"
"Pid: %d\n"
"Uptime: %dd %dh%02dm%02ds\n"
"Uptime_sec: %d\n"
"Memmax_MB: %d\n"
"Ulimit-n: %d\n"
"Maxsock: %d\n"
"Maxconn: %d\n"
[MEDIUM] stats: add the "set maxconn" setting to the command line interface This option permits to change the global maxconn setting within the limit that was set by the initial value, which is now reported as the hard maxconn value. This allows to immediately accept more concurrent connections or to stop accepting new ones until the value passes below the indicated setting. The main use of this option is on systems where many haproxy instances are loaded and admins need to re-adjust resource sharing at run time to regain a bit of fairness between processes.
2011-09-07 08:38:31 -04:00
"Hard_maxconn: %d\n"
[STATS] report pipe usage in the statistics Pipe usage is reported in info and web stats including maxpipes, pipes_free, and pipes_used.
2009-01-25 08:02:00 -05:00
"Maxpipes: %d\n"
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
"CurrConns: %d\n"
[STATS] report pipe usage in the statistics Pipe usage is reported in info and web stats including maxpipes, pipes_free, and pipes_used.
2009-01-25 08:02:00 -05:00
"PipesUsed: %d\n"
"PipesFree: %d\n"
[MINOR] stats: report the current and max global connection rates The HTML page reports the current process connection rate, and the "show info" command on the stats socket also reports the conn rate limit and the max conn rate that was once reached. Note that the max value can be cleared using "clear counters".
2011-09-07 09:26:48 -04:00
"ConnRate: %d\n"
"ConnRateLimit: %d\n"
"MaxConnRate: %d\n"
[MINOR] stats: report number of tasks (active and running) It may be useful for statistics purposes to report the number of tasks.
2009-03-21 13:33:52 -04:00
"Tasks: %d\n"
"Run_queue: %d\n"
[MEDIUM] poll: add a measurement of idle vs work time We now measure the work and idle times in order to report the idle time in the stats. It's expected that we'll be able to use it at other places later.
2011-09-10 10:56:42 -04:00
"Idle_pct: %d\n"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"node: %s\n"
"description: %s\n"
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
"",
global.nbproc,
relative_pid,
pid,
up / 86400, (up % 86400) / 3600, (up % 3600) / 60, (up % 60),
up,
global.rlimit_memmax,
global.rlimit_nofile,
[MEDIUM] stats: add the "set maxconn" setting to the command line interface This option permits to change the global maxconn setting within the limit that was set by the initial value, which is now reported as the hard maxconn value. This allows to immediately accept more concurrent connections or to stop accepting new ones until the value passes below the indicated setting. The main use of this option is on systems where many haproxy instances are loaded and admins need to re-adjust resource sharing at run time to regain a bit of fairness between processes.
2011-09-07 08:38:31 -04:00
global.maxsock, global.maxconn, global.hardmaxconn, global.maxpipes,
[MINOR] stats: report number of tasks (active and running) It may be useful for statistics purposes to report the number of tasks.
2009-03-21 13:33:52 -04:00
actconn, pipes_used, pipes_free,
[MINOR] stats: report the current and max global connection rates The HTML page reports the current process connection rate, and the "show info" command on the stats socket also reports the conn rate limit and the max conn rate that was once reached. Note that the max value can be cleared using "clear counters".
2011-09-07 09:26:48 -04:00
read_freq_ctr(&global.conn_per_sec), global.cps_lim, global.cps_max,
[MEDIUM] poll: add a measurement of idle vs work time We now measure the work and idle times in order to report the idle time in the stats. It's expected that we'll be able to use it at other places later.
2011-09-10 10:56:42 -04:00
nb_tasks_cur, run_queue_cur, idle_pct,
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
global.node, global.desc?global.desc:""
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
return 0;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px = proxy;
si->applet.ctx.stats.px_st = STAT_PX_ST_INIT;
si->applet.ctx.stats.sv = NULL;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_LIST;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_LIST:
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
/* dump proxies */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_SHOW_STAT) {
while (si->applet.ctx.stats.px) {
px = si->applet.ctx.stats.px;
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
/* skip the disabled proxies and non-networked ones */
if (px->state != PR_STSTOPPED &&
[MINOR] stats: make stats_dump_raw_to_buffer() use buffer_feed_chunk Same as previous change. A remaining call to stats_dump_proxy() still prevents us from completing the update.
2009-10-04 05:54:04 -04:00
(px->cap & (PR_CAP_FE | PR_CAP_BE))) {
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_proxy(si, px, NULL) == 0)
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
return 0;
[MINOR] stats: make stats_dump_raw_to_buffer() use buffer_feed_chunk Same as previous change. A remaining call to stats_dump_proxy() still prevents us from completing the update.
2009-10-04 05:54:04 -04:00
}
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px = px->next;
si->applet.ctx.stats.px_st = STAT_PX_ST_INIT;
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
}
/* here, we just have reached the last proxy */
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
}
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_END;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_END:
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MINOR] slightly rebalance stats_dump_{raw,http} Both should process the response buffer equally. They now both clear the hijack bit once done, and both receive a pointer to the response buffer in their arguments.
2008-12-07 12:30:00 -05:00
/* fall through */
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_FIN:
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
return 1;
default:
/* unknown state ! */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MINOR] stats: make stats_dump_raw_to_buffer() use buffer_feed_chunk Same as previous change. A remaining call to stats_dump_proxy() still prevents us from completing the update.
2009-10-04 05:54:04 -04:00
return 1;
[MEDIUM] implement the statistics output on a unix socket A unix socket can now access the statistics. It currently only recognizes the "show stat\n" command at the beginning of the input, then returns the statistics in CSV format.
2007-10-17 12:57:38 -04:00
}
}
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* We don't want to land on the posted stats page because a refresh will
* repost the data. We don't want this to happen on accident so we redirect
* the browse to the stats page with a GET.
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_http_redir(struct stream_interface *si, struct uri_auth *uri)
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
struct chunk msg;
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
switch (si->conn.data_st) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INIT:
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg,
"HTTP/1.0 303 See Other\r\n"
"Cache-Control: no-cache\r\n"
"Content-Type: text/plain\r\n"
"Connection: close\r\n"
"Location: %s;st=%s",
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
uri->uri_prefix,
((si->applet.ctx.stats.st_code > STAT_STATUS_INIT) &&
(si->applet.ctx.stats.st_code < STAT_STATUS_SIZE) &&
stat_status_codes[si->applet.ctx.stats.st_code]) ?
stat_status_codes[si->applet.ctx.stats.st_code] :
stat_status_codes[STAT_STATUS_UNKN]);
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg, "\r\n\r\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
return 0;
s->txn.status = 303;
if (!(s->flags & SN_ERR_MASK)) // this is not really an error but it is
s->flags |= SN_ERR_PRXCOND; // to mark that it comes from the proxy
if (!(s->flags & SN_FINST_MASK))
s->flags |= SN_FINST_R;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
return 1;
}
return 1;
}
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
/* This I/O handler runs as an applet embedded in a stream interface. It is
* used to send HTTP stats over a TCP socket. The mechanism is very simple.
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
* si->applet.st0 becomes non-zero once the transfer is finished. The handler
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
* automatically unregisters itself once transfer is complete.
*/
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
static void http_stats_io_handler(struct stream_interface *si)
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
REORG/MAJOR: use "struct channel" instead of "struct buffer" This is a massive rename. We'll then split channel and buffer. This change needs a lot of cleanups. At many locations, the parameter or variable is still called "buf" which will become ambiguous. Also, the "struct channel" is still defined in buffers.h.
2012-07-02 09:11:27 -04:00
struct channel *req = si->ob;
struct channel *res = si->ib;
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
if (unlikely(si->state == SI_ST_DIS || si->state == SI_ST_CLO))
goto out;
/* check that the output is not closed */
if (res->flags & (BF_SHUTW|BF_SHUTW_NOW))
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = 1;
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if (!si->applet.st0) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
if (s->txn.meth == HTTP_METH_POST) {
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_http_redir(si, s->be->uri_auth)) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = 1;
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutw(si);
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
}
} else {
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_http(si, s->be->uri_auth)) {
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
si->applet.st0 = 1;
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutw(si);
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
}
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
}
}
if ((res->flags & BF_SHUTR) && (si->state == SI_ST_EST))
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutw(si);
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
[REORG] stream_interface: move the st0, st1 and private members to the applet Those fields are only used by the applets, so let's move them to the struct.
2011-02-13 07:25:14 -05:00
if ((req->flags & BF_SHUTW) && (si->state == SI_ST_EST) && si->applet.st0) {
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_shutr(si);
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
res->flags |= BF_READ_NULL;
}
/* update all other flags and resync with the other side */
MINOR: stream_interface: introduce a new "struct connection" type We start to move everything needed to manage a connection to a special entity "struct connection". We have the data layer operations and the control operations there. We'll also have more info in the future such as file descriptors and applet contexts, so that in the end it becomes detachable from the stream interface, which will allow connections to be reused between sessions. For now on, we start with minimal changes.
2012-05-21 10:31:45 -04:00
si_update(si);
[MEDIUM] stats: make HTTP stats use an I/O handler Doing this, we can remove the last BF_HIJACK user and remove produce_content(). s->data_source could also be removed but it is currently used to detect if the stats or a server was used.
2009-10-04 09:56:38 -04:00
/* we don't want to expire timeouts while we're processing requests */
si->ib->rex = TICK_ETERNITY;
si->ob->wex = TICK_ETERNITY;
out:
if (unlikely(si->state == SI_ST_DIS || si->state == SI_ST_CLO)) {
/* check that we have released everything then unregister */
stream_int_unregister_handler(si);
}
}
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps statistics in HTTP format onto the stream interface's
* read buffer. The data_ctx must have been zeroed first, and the flags
* properly set. It returns 0 if it had to stop writing data and an I/O is
* needed, 1 if the dump is finished and the session must be closed, or -1
* in case of any error.
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_http(struct stream_interface *si, struct uri_auth *uri)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
REORG/MAJOR: use "struct channel" instead of "struct buffer" This is a massive rename. We'll then split channel and buffer. This change needs a lot of cleanups. At many locations, the parameter or variable is still called "buf" which will become ambiguous. Also, the "struct channel" is still defined in buffers.h.
2012-07-02 09:11:27 -04:00
struct channel *rep = si->ib;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
struct proxy *px;
struct chunk msg;
unsigned int up;
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
switch (si->conn.data_st) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INIT:
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"HTTP/1.0 200 OK\r\n"
"Cache-Control: no-cache\r\n"
"Connection: close\r\n"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"Content-Type: %s\r\n",
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_FMT_CSV) ? "text/plain" : "text/html");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (uri->refresh > 0 && !(si->applet.ctx.stats.flags & STAT_NO_REFRESH))
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "Refresh: %d\r\n",
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
uri->refresh);
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "\r\n");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
s->txn.status = 200;
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] stats: prepare the connection for closing before dumping We will need to modify the stats dump functions so that they can be used in interactive mode. For this, we want their caller to prepare the connection for a close, not themselves to do it. Let's simply move the stream_int_retnclose() out.
2009-09-22 13:27:35 -04:00
return 0;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
if (!(s->flags & SN_ERR_MASK)) // this is not really an error but it is
s->flags |= SN_ERR_PRXCOND; // to mark that it comes from the proxy
if (!(s->flags & SN_FINST_MASK))
s->flags |= SN_FINST_R;
if (s->txn.meth == HTTP_METH_HEAD) {
/* that's all we return in case of HEAD request */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 1;
}
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_HEAD; /* let's start producing data */
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_HEAD:
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
/* WARNING! This must fit in the first buffer !!! */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"\n"
"\"http://www.w3.org/TR/html4/loose.dtd\">\n"
[MINOR] stats: add a new node-name setting The new "node-name" stats setting enables reporting of a node ID on the stats page. It is possible to return the system's host name as well as a specific name.
2009-08-16 04:29:18 -04:00
"<html><head><title>Statistics Report for " PRODUCT_NAME "%s%s</title>\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<meta http-equiv=\"content-type\" content=\"text/html; charset=iso-8859-1\">\n"
"<style type=\"text/css\"><!--\n"
"body {"
[MINOR] stats/html: use the arial font before helvetica The stats HTML output were barely readable on some browsers such as firefox on Linux, due to the selected helvetica font which is too small. Specifying "arial" first fixes the issue without changing the table size. Also, the default size of 0.8em choosen to get 10px out of 12px is wrong because it gets 9px when rounded down.
2009-05-10 14:08:10 -04:00
" font-family: arial, helvetica, sans-serif;"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
" font-size: 12px;"
" font-weight: normal;"
" color: black;"
" background: white;"
"}\n"
"th,td {"
[MINOR] stats/html: use the arial font before helvetica The stats HTML output were barely readable on some browsers such as firefox on Linux, due to the selected helvetica font which is too small. Specifying "arial" first fixes the issue without changing the table size. Also, the default size of 0.8em choosen to get 10px out of 12px is wrong because it gets 9px when rounded down.
2009-05-10 14:08:10 -04:00
" font-size: 10px;"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"}\n"
"h1 {"
[MINOR] stats: better displaying in MSIE MSIE does not correctly display spaced digits. It requires a margin of at least one pixel. Also, it does not correctly hide empty cells, so we work around this by setting the background white. Last, the H1 font was too large, so we reduce it by one size, which is still OK in other browsers.
2009-07-15 04:07:05 -04:00
" font-size: x-large;"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
" margin-bottom: 0.5em;"
"}\n"
"h2 {"
" font-family: helvetica, arial;"
" font-size: x-large;"
" font-weight: bold;"
" font-style: italic;"
" color: #6020a0;"
" margin-top: 0em;"
" margin-bottom: 0em;"
"}\n"
"h3 {"
" font-family: helvetica, arial;"
" font-size: 16px;"
" font-weight: bold;"
" color: #b00040;"
" background: #e8e8d0;"
" margin-top: 0em;"
" margin-bottom: 0em;"
"}\n"
"li {"
" margin-top: 0.25em;"
" margin-right: 2em;"
"}\n"
".hr {margin-top: 0.25em;"
" border-color: black;"
" border-bottom-style: solid;"
"}\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
".titre {background: #20D0D0;color: #000000; font-weight: bold; text-align: center;}\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
".total {background: #20D0D0;color: #ffff80;}\n"
".frontend {background: #e8e8d0;}\n"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
".socket {background: #d0d0d0;}\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
".backend {background: #e8e8d0;}\n"
".active0 {background: #ff9090;}\n"
".active1 {background: #ffd020;}\n"
".active2 {background: #ffffa0;}\n"
".active3 {background: #c0ffc0;}\n"
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
".active4 {background: #ffffa0;}\n" /* NOLB state shows same as going down */
".active5 {background: #a0e0a0;}\n" /* NOLB state shows darker than up */
".active6 {background: #e0e0e0;}\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
".backup0 {background: #ff9090;}\n"
".backup1 {background: #ff80ff;}\n"
".backup2 {background: #c060ff;}\n"
".backup3 {background: #b0d0ff;}\n"
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
".backup4 {background: #c060ff;}\n" /* NOLB state shows same as going down */
".backup5 {background: #90b0e0;}\n" /* NOLB state shows same as going down */
".backup6 {background: #e0e0e0;}\n"
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
".maintain {background: #c07820;}\n"
[MINOR] stats: better displaying in MSIE MSIE does not correctly display spaced digits. It requires a margin of at least one pixel. Also, it does not correctly hide empty cells, so we work around this by setting the background white. Last, the H1 font was too large, so we reduce it by one size, which is still OK in other browsers.
2009-07-15 04:07:05 -04:00
".rls {letter-spacing: 0.2em; margin-right: 1px;}\n" /* right letter spacing (used for grouping digits) */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"\n"
[MINOR] add additional "a href"s to stats page This patch adds <a href> html links for proxies, frontends, servers and backends. Once located, can be clicked. Users no longer have to manually add #anchor to stat's url.
2009-10-24 08:24:30 -04:00
"a.px:link {color: #ffff40; text-decoration: none;}"
"a.px:visited {color: #ffff40; text-decoration: none;}"
"a.px:hover {color: #ffffff; text-decoration: none;}"
"a.lfsb:link {color: #000000; text-decoration: none;}"
"a.lfsb:visited {color: #000000; text-decoration: none;}"
"a.lfsb:hover {color: #505050; text-decoration: none;}"
"\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"table.tbl { border-collapse: collapse; border-style: none;}\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"table.tbl td { text-align: right; border-width: 1px 1px 1px 1px; border-style: solid solid solid solid; padding: 2px 3px; border-color: gray; white-space: nowrap;}\n"
"table.tbl td.ac { text-align: center;}\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"table.tbl th { border-width: 1px; border-style: solid solid solid solid; border-color: gray;}\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"table.tbl th.pxname { background: #b00040; color: #ffff40; font-weight: bold; border-style: solid solid none solid; padding: 2px 3px; white-space: nowrap;}\n"
[MINOR] stats: better displaying in MSIE MSIE does not correctly display spaced digits. It requires a margin of at least one pixel. Also, it does not correctly hide empty cells, so we work around this by setting the background white. Last, the H1 font was too large, so we reduce it by one size, which is still OK in other browsers.
2009-07-15 04:07:05 -04:00
"table.tbl th.empty { border-style: none; empty-cells: hide; background: white;}\n"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"table.tbl th.desc { background: white; border-style: solid solid none solid; text-align: left; padding: 2px 3px;}\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"table.lgd { border-collapse: collapse; border-width: 1px; border-style: none none none solid; border-color: black;}\n"
"table.lgd td { border-width: 1px; border-style: solid solid solid solid; border-color: gray; padding: 2px;}\n"
"table.lgd td.noborder { border-style: none; padding: 2px; white-space: nowrap;}\n"
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
"u {text-decoration:none; border-bottom: 1px dotted black;}\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"-->\n"
[MINOR] stats: add a new node-name setting The new "node-name" stats setting enables reporting of a node ID on the stats page. It is possible to return the system's host name as well as a specific name.
2009-08-16 04:29:18 -04:00
"</style></head>\n",
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
(uri->flags&ST_SHNODE) ? " on " : "",
(uri->flags&ST_SHNODE) ? (uri->node ? uri->node : global.node) : ""
[MINOR] stats: add a new node-name setting The new "node-name" stats setting enables reporting of a node ID on the stats page. It is possible to return the system's host name as well as a specific name.
2009-08-16 04:29:18 -04:00
);
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
} else {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
print_csv_header(&msg);
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
}
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 0;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INFO;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INFO:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
up = (now.tv_sec - start_date.tv_sec);
/* WARNING! this has to fit the first packet too.
* We are around 3.5 kB, add adding entries will
* become tricky if we want to support 4kB buffers !
*/
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<body><h1><a href=\"" PRODUCT_URL "\" style=\"text-decoration: none;\">"
PRODUCT_NAME "%s</a></h1>\n"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"<h2>Statistics Report for pid %d%s%s%s%s</h2>\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<hr width=\"100%%\" class=\"hr\">\n"
"<h3>&gt; General process information</h3>\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<table border=0><tr><td align=\"left\" nowrap width=\"1%%\">\n"
[STATS] add support for "show info" on the unix socket It is sometimes required to know some informations such as the process uptime when consulting statistics. This patch adds the "show info" command to query those informations on the UNIX socket.
2008-01-03 04:19:15 -05:00
"<p><b>pid = </b> %d (process #%d, nbproc = %d)<br>\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<b>uptime = </b> %dd %dh%02dm%02ds<br>\n"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"<b>system limits:</b> memmax = %s%s; ulimit-n = %d<br>\n"
"<b>maxsock = </b> %d; <b>maxconn = </b> %d; <b>maxpipes = </b> %d<br>\n"
[MINOR] stats: report the current and max global connection rates The HTML page reports the current process connection rate, and the "show info" command on the stats socket also reports the conn rate limit and the max conn rate that was once reached. Note that the max value can be cleared using "clear counters".
2011-09-07 09:26:48 -04:00
"current conns = %d; current pipes = %d/%d; conn rate = %d/sec<br>\n"
[MEDIUM] poll: add a measurement of idle vs work time We now measure the work and idle times in order to report the idle time in the stats. It's expected that we'll be able to use it at other places later.
2011-09-10 10:56:42 -04:00
"Running tasks: %d/%d; idle = %d %%<br>\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"</td><td align=\"center\" nowrap>\n"
"<table class=\"lgd\"><tr>\n"
"<td class=\"active3\">&nbsp;</td><td class=\"noborder\">active UP </td>"
"<td class=\"backup3\">&nbsp;</td><td class=\"noborder\">backup UP </td>"
"</tr><tr>\n"
"<td class=\"active2\"></td><td class=\"noborder\">active UP, going down </td>"
"<td class=\"backup2\"></td><td class=\"noborder\">backup UP, going down </td>"
"</tr><tr>\n"
"<td class=\"active1\"></td><td class=\"noborder\">active DOWN, going up </td>"
"<td class=\"backup1\"></td><td class=\"noborder\">backup DOWN, going up </td>"
"</tr><tr>\n"
"<td class=\"active0\"></td><td class=\"noborder\">active or backup DOWN &nbsp;</td>"
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
"<td class=\"active6\"></td><td class=\"noborder\">not checked </td>"
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
"</tr><tr>\n"
"<td class=\"maintain\"></td><td class=\"noborder\" colspan=\"3\">active or backup DOWN for maintenance (MAINT) &nbsp;</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"</tr></table>\n"
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
"Note: UP with load-balancing disabled is reported as \"NOLB\"."
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"</td>"
"<td align=\"left\" valign=\"top\" nowrap width=\"1%%\">"
"<b>Display option:</b><ul style=\"margin-top: 0.25em;\">"
"",
(uri->flags&ST_HIDEVER)?"":(STATS_VERSION_STRING),
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
pid, (uri->flags&ST_SHNODE) ? " on " : "", (uri->flags&ST_SHNODE) ? (uri->node ? uri->node : global.node) : "",
(uri->flags&ST_SHDESC)? ": " : "", (uri->flags&ST_SHDESC) ? (uri->desc ? uri->desc : global.desc) : "",
pid, relative_pid, global.nbproc,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
up / 86400, (up % 86400) / 3600,
(up % 3600) / 60, (up % 60),
global.rlimit_memmax ? ultoa(global.rlimit_memmax) : "unlimited",
global.rlimit_memmax ? " MB" : "",
global.rlimit_nofile,
[STATS] report pipe usage in the statistics Pipe usage is reported in info and web stats including maxpipes, pipes_free, and pipes_used.
2009-01-25 08:02:00 -05:00
global.maxsock, global.maxconn, global.maxpipes,
[MINOR] stats: report the current and max global connection rates The HTML page reports the current process connection rate, and the "show info" command on the stats socket also reports the conn rate limit and the max conn rate that was once reached. Note that the max value can be cleared using "clear counters".
2011-09-07 09:26:48 -04:00
actconn, pipes_used, pipes_used+pipes_free, read_freq_ctr(&global.conn_per_sec),
[MEDIUM] poll: add a measurement of idle vs work time We now measure the work and idle times in order to report the idle time in the stats. It's expected that we'll be able to use it at other places later.
2011-09-10 10:56:42 -04:00
run_queue_cur, nb_tasks_cur, idle_pct
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
);
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_HIDE_DOWN)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"%s%s%s\">Show all servers</a><br>\n",
uri->uri_prefix,
"",
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"%s%s%s\">Hide 'DOWN' servers</a><br>\n",
uri->uri_prefix,
";up",
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
if (uri->refresh > 0) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_NO_REFRESH)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"%s%s%s\">Enable refresh</a><br>\n",
uri->uri_prefix,
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_HIDE_DOWN) ? ";up" : "",
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"%s%s%s\">Disable refresh</a><br>\n",
uri->uri_prefix,
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_HIDE_DOWN) ? ";up" : "",
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
";norefresh");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"%s%s%s\">Refresh now</a><br>\n",
uri->uri_prefix,
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(si->applet.ctx.stats.flags & STAT_HIDE_DOWN) ? ";up" : "",
(si->applet.ctx.stats.flags & STAT_NO_REFRESH) ? ";norefresh" : "");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] add a link to the CSV export on the stats page.
2007-10-18 05:05:48 -04:00
"<li><a href=\"%s;csv%s\">CSV export</a><br>\n",
uri->uri_prefix,
(uri->refresh > 0) ? ";norefresh" : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"</ul></td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<td align=\"left\" valign=\"top\" nowrap width=\"1%%\">"
MINOR: patch for minor typo (ressources/resources) The main stats page says "ressources" (French spelling) rather than "resources" (English spelling). One little patch attached (against v1.4.20). Many thanks, Adrian
2012-03-19 19:36:42 -04:00
"<b>External resources:</b><ul style=\"margin-top: 0.25em;\">\n"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"<li><a href=\"" PRODUCT_URL "\">Primary site</a><br>\n"
"<li><a href=\"" PRODUCT_URL_UPD "\">Updates (v" PRODUCT_BRANCH ")</a><br>\n"
"<li><a href=\"" PRODUCT_URL_DOC "\">Online manual</a><br>\n"
"</ul>"
"</td>"
"</tr></table>\n"
""
);
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.st_code) {
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
switch (si->applet.ctx.stats.st_code) {
case STAT_STATUS_DONE:
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg,
"<p><div class=active3>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"Action processed successfully."
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
case STAT_STATUS_NONE:
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg,
"<p><div class=active2>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"Nothing has changed."
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
case STAT_STATUS_PART:
MINOR: stats admin: allow unordered parameters in POST requests Previously, the stats admin page required POST parameters to be provided exactly in the same order as the HTML form. This patch allows to handle those parameters in any orders. Also, note that haproxy won't alter server states anymore if backend or server names are ambiguous (duplicated names in the configuration) to prevent unexpected results (the same should probably be applied to the stats socket).
2012-04-04 06:57:18 -04:00
chunk_printf(&msg,
"<p><div class=active2>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"Action partially processed.<br>"
"Some server names are probably unknown or ambiguous (duplicated names in the backend)."
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
case STAT_STATUS_ERRP:
MINOR: stats admin: allow unordered parameters in POST requests Previously, the stats admin page required POST parameters to be provided exactly in the same order as the HTML form. This patch allows to handle those parameters in any orders. Also, note that haproxy won't alter server states anymore if backend or server names are ambiguous (duplicated names in the configuration) to prevent unexpected results (the same should probably be applied to the stats socket).
2012-04-04 06:57:18 -04:00
chunk_printf(&msg,
"<p><div class=active0>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"Action not processed because of invalid parameters."
"<ul>"
"<li>The action is maybe unknown.</li>"
"<li>The backend name is probably unknown or ambiguous (duplicated names).</li>"
"<li>Some server names are probably unknown or ambiguous (duplicated names in the backend).</li>"
"</ul>"
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
case STAT_STATUS_EXCD:
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg,
"<p><div class=active0>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"<b>Action not processed : the buffer couldn't store all the data.<br>"
"You should retry with less servers at a time.</b>"
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
case STAT_STATUS_DENY:
[MEDIUM] stats: add an admin level The stats web interface must be read-only by default to prevent security holes. As it is now allowed to enable/disable servers, a new keyword "stats admin" is introduced to activate this admin level, conditioned by ACLs. (cherry picked from commit 5334bab92ca7debe36df69983c19c21b6dc63f78)
2010-10-11 18:14:36 -04:00
chunk_printf(&msg,
"<p><div class=active0>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"<b>Action denied.</b>"
"</div>\n", uri->uri_prefix);
MINOR: stats admin: reduce memcmp()/strcmp() calls on status codes memcmp()/strcmp() calls were needed in different parts of code to determine the status code. Each new status code introduces new calls, which can become inefficient and source of bugs. This patch reorganizes the code to rely on a numeric status code internally and to be hopefully more generic.
2012-04-04 06:57:21 -04:00
break;
default:
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg,
"<p><div class=active6>"
"<a class=lfsb href=\"%s\" title=\"Remove this message\">[X]</a> "
"Unexpected result."
"</div>\n", uri->uri_prefix);
}
chunk_printf(&msg,"<p>\n");
}
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
return 0;
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px = proxy;
si->applet.ctx.stats.px_st = STAT_PX_ST_INIT;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_LIST;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_LIST:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* dump proxies */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
while (si->applet.ctx.stats.px) {
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
if (buffer_almost_full(&rep->buf))
[OPTIM] stats: check free space before trying to print This alone makes a typical HTML stats dump consume 10% CPU less, because we avoid doing complex printf calls to drop them later. Only a few common cases have been checked, those which are very likely to run for nothing.
2009-10-11 17:35:10 -04:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
px = si->applet.ctx.stats.px;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* skip the disabled proxies and non-networked ones */
if (px->state != PR_STSTOPPED && (px->cap & (PR_CAP_FE | PR_CAP_BE)))
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (stats_dump_proxy(si, px, uri) == 0)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px = px->next;
si->applet.ctx.stats.px_st = STAT_PX_ST_INIT;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
}
/* here, we just have reached the last proxy */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_END;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_END:
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "</body></html>\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
return 0;
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_FIN:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 1;
default:
/* unknown state ! */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return -1;
}
}
/*
* Dumps statistics for a proxy.
* Returns 0 if it had to stop dumping data because of lack of buffer space,
* ot non-zero if everything completed.
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_proxy(struct stream_interface *si, struct proxy *px, struct uri_auth *uri)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
REORG/MAJOR: use "struct channel" instead of "struct buffer" This is a massive rename. We'll then split channel and buffer. This change needs a lot of cleanups. At many locations, the parameter or variable is still called "buf" which will become ambiguous. Also, the "struct channel" is still defined in buffers.h.
2012-07-02 09:11:27 -04:00
struct channel *rep = si->ib;
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
struct server *sv, *svs; /* server and server-state, server-state=server or server->track */
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
struct listener *l;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
struct chunk msg;
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
switch (si->applet.ctx.stats.px_st) {
case STAT_PX_ST_INIT:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* we are on a new proxy */
if (uri && uri->scope) {
/* we have a limited scope, we have to check the proxy name */
struct stat_scope *scope;
int len;
len = strlen(px->id);
scope = uri->scope;
while (scope) {
/* match exact proxy name */
if (scope->px_len == len && !memcmp(px->id, scope->px_id, len))
break;
/* match '.' which means 'self' proxy */
[BUG] scope "." must match the backend and not the frontend
2007-10-18 10:38:37 -04:00
if (!strcmp(scope->px_id, ".") && px == s->be)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
break;
scope = scope->next;
}
/* proxy name not found : don't dump anything */
if (scope == NULL)
return 1;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if ((si->applet.ctx.stats.flags & STAT_BOUND) && (si->applet.ctx.stats.iid != -1) &&
(px->uuid != si->applet.ctx.stats.iid))
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
return 1;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px_st = STAT_PX_ST_TH;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_TH:
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* A form to enable/disable this proxy servers */
chunk_printf(&msg,
"<form action=\"%s\" method=\"post\">",
uri->uri_prefix);
}
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* print a new table */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"<table class=\"tbl\" width=\"100%%\">\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<tr class=\"titre\">"
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
"<th class=\"pxname\" width=\"10%%\"");
if (uri->flags&ST_SHLGNDS) {
/* cap, mode, id */
chunk_printf(&msg, " title=\"cap: %s, mode: %s, id: %d",
proxy_cap_str(px->cap), proxy_mode_str(px->mode),
px->uuid);
chunk_printf(&msg, "\"");
}
chunk_printf(&msg,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s<a name=\"%s\"></a>"
"<a class=px href=\"#%s\">%s</a>%s</th>"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"<th class=\"%s\" width=\"90%%\">%s</th>"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"</tr>\n"
[MINOR] add "description", "node" and show-node"/"show-desc", remove "node-name", v2 This patch implements "description" (proxy and global) and "node" (global) options, removes "node-name" and adds "show-node" & "show-desc" options for "stats". It also changes the way the header lines (with proxy name) and the statistics are displayed, so stats no longer look so clumsy with very long names. Instead of "node-name" it is possible to use show-node/show-desc with an optional parameter that overrides a default node/description. backend cust-0045 # report specific values for this customer stats show-node Europe stats show-desc Master node for Europe, Asia, Africa
2009-10-02 16:51:14 -04:00
"</table>\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<table class=\"tbl\" width=\"100%%\">\n"
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"<tr class=\"titre\">",
(uri->flags & ST_SHLGNDS)?"<u>":"",
px->id, px->id, px->id,
(uri->flags & ST_SHLGNDS)?"</u>":"",
px->desc ? "desc" : "empty", px->desc ? px->desc : "");
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* Column heading for Enable or Disable server */
chunk_printf(&msg, "<th rowspan=2 width=1></th>");
}
chunk_printf(&msg,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"<th rowspan=2></th>"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
"<th colspan=3>Queue</th>"
"<th colspan=3>Session rate</th><th colspan=5>Sessions</th>"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"<th colspan=2>Bytes</th><th colspan=2>Denied</th>"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
"<th colspan=3>Errors</th><th colspan=2>Warnings</th>"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
"<th colspan=9>Server</th>"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"</tr>\n"
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<tr class=\"titre\">"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
"<th>Cur</th><th>Max</th><th>Limit</th>"
[MEDIUM] add support for "maxqueue" to limit server queue overload This patch adds the "maxqueue" parameter to the server. This allows new sessions to be immediately rebalanced when the server's queue is filled. It's useful when session stickiness is just a performance boost (even a huge one) but not a requirement. This should only be used if session affinity isn't a hard functional requirement but provides performance boost by keeping server-local caches hot and compact). Absence of 'maxqueue' option means unlimited queue. When queue gets filled up to 'maxqueue' client session is moved from server-local queue to a global one.
2007-10-25 14:15:38 -04:00
"<th>Cur</th><th>Max</th><th>Limit</th><th>Cur</th><th>Max</th>"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
"<th>Limit</th><th>Total</th><th>LbTot</th><th>In</th><th>Out</th>"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"<th>Req</th><th>Resp</th><th>Req</th><th>Conn</th>"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
"<th>Resp</th><th>Retr</th><th>Redis</th>"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
"<th>Status</th><th>LastChk</th><th>Wght</th><th>Act</th>"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"<th>Bck</th><th>Chk</th><th>Dwn</th><th>Dwntme</th>"
"<th>Thrtle</th>\n"
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"</tr>");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
return 0;
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px_st = STAT_PX_ST_FE;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_FE:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* print the frontend */
[MEDIUM] fix stats socket limitation to 16 kB Due to the way the stats socket work, it was not possible to maintain the information related to the command entered, so after filling a whole buffer, the request was lost and it was considered that there was nothing to write anymore. The major reason was that some flags were passed directly during the first call to stats_dump_raw() instead of being stored persistently in the session. To definitely fix this problem, flags were added to the stats member of the session structure. A second problem appeared. When the stats were produced, a first call to client_retnclose() was performed, then one or multiple subsequent calls to buffer_write_chunks() were done. But once the stats buffer was full and a reschedule operated, the buffer was flushed, the write flag cleared from the buffer and nothing was done to re-arm it. For this reason, a check was added in the proto_uxst_stats() function in order to re-call the client FSM when data were added by stats_dump_raw(). Finally, the whole unix stats dump FSM was rewritten to avoid all the magics it depended on. It is now simpler and looks more like the HTTP one.
2008-03-17 16:38:24 -04:00
if ((px->cap & PR_CAP_FE) &&
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(!(si->applet.ctx.stats.flags & STAT_BOUND) || (si->applet.ctx.stats.type & (1 << STATS_TYPE_FE)))) {
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* name, queue */
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"<tr class=\"frontend\">");
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* Column sub-heading for Enable or Disable server */
chunk_printf(&msg, "<td></td>");
}
chunk_printf(&msg,
"<td class=ac>"
[MINOR] add additional "a href"s to stats page This patch adds <a href> html links for proxies, frontends, servers and backends. Once located, can be clicked. Users no longer have to manually add #anchor to stat's url.
2009-10-24 08:24:30 -04:00
"<a name=\"%s/Frontend\"></a>"
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"<a class=lfsb href=\"#%s/Frontend\">Frontend</a></td>"
"<td colspan=3></td>"
[MINOR] stats: report the request rate in frontends as cell titles The current and max request rates are now reported when the mouse flies over the session rate cur/max. The total requests is displayed with the status codes over the total sessions cell.
2010-02-26 05:35:39 -05:00
"",
px->id, px->id);
if (px->mode == PR_MODE_HTTP) {
chunk_printf(&msg,
/* sessions rate : current, max, limit */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
"<td title=\"Cur: %u req/s\"><u>%s</u></td><td title=\"Max: %u req/s\"><u>%s</u></td><td>%s</td>"
[MINOR] stats: report the request rate in frontends as cell titles The current and max request rates are now reported when the mouse flies over the session rate cur/max. The total requests is displayed with the status codes over the total sessions cell.
2010-02-26 05:35:39 -05:00
"",
read_freq_ctr(&px->fe_req_per_sec),
U2H0(read_freq_ctr(&px->fe_sess_per_sec)),
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->fe_counters.p.http.rps_max,
U2H1(px->fe_counters.sps_max),
[MINOR] stats: report the request rate in frontends as cell titles The current and max request rates are now reported when the mouse flies over the session rate cur/max. The total requests is displayed with the status codes over the total sessions cell.
2010-02-26 05:35:39 -05:00
LIM2A2(px->fe_sps_lim, "-"));
} else {
chunk_printf(&msg,
/* sessions rate : current, max, limit */
"<td>%s</td><td>%s</td><td>%s</td>"
"",
U2H0(read_freq_ctr(&px->fe_sess_per_sec)),
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H1(px->fe_counters.sps_max), LIM2A2(px->fe_sps_lim, "-"));
[MINOR] stats: report the request rate in frontends as cell titles The current and max request rates are now reported when the mouse flies over the session rate cur/max. The total requests is displayed with the status codes over the total sessions cell.
2010-02-26 05:35:39 -05:00
}
chunk_printf(&msg,
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
/* sessions: current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td>%s</td>"
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
"<td"
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H3(px->feconn), U2H4(px->fe_counters.conn_max), U2H5(px->maxconn));
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
/* http response (via td title): 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, " title=\"%lld requests:", px->fe_counters.p.http.cum_req);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
for (i = 1; i < 6; i++)
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, " %dxx=%lld,", i, px->fe_counters.p.http.rsp[i]);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
[MINOR] stats: report the number of requests intercepted by the frontend These requests are mainly monitor requests, as well as stats requests when the stats are processed by the frontend. Having this counter helps explain the difference in number of sessions that is sometimes observed between a frontend and a backend.
2011-09-10 17:29:44 -04:00
chunk_printf(&msg, " other=%lld,", px->fe_counters.p.http.rsp[0]);
chunk_printf(&msg, " intercepted=%lld\"", px->fe_counters.intercepted_req);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
}
chunk_printf(&msg,
/* sessions: total, lbtot */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s%s%s</td><td></td>"
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
/* bytes : in, out */
"<td>%s</td><td>%s</td>"
"",
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(px->mode == PR_MODE_HTTP)?"<u>":"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H6(px->fe_counters.cum_sess),
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(px->mode == PR_MODE_HTTP)?"</u>":"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H7(px->fe_counters.bytes_in), U2H8(px->fe_counters.bytes_out));
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* denied: req, resp */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* errors : request, connect, response */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td></td><td></td>"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td></td><td></td>"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* server status : reflect frontend status */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>%s</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* rest of server: nothing */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac colspan=8></td></tr>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H0(px->fe_counters.denied_req), U2H1(px->fe_counters.denied_resp),
U2H2(px->fe_counters.failed_req),
[CLEANUP] proxy: rename a few proxy states (PR_STIDLE and PR_STRUN) Those states have been replaced with PR_STFULL and PR_STREADY respectively, as it is what matches them the best now. Also, two occurrences of PR_STIDLE in peers.c have been removed as this did not provide any form of error recovery anyway.
2011-07-25 02:11:52 -04:00
px->state == PR_STREADY ? "OPEN" :
px->state == PR_STFULL ? "FULL" : "STOP");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
} else {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* pxid, name, queue cur, queue max, */
"%s,FRONTEND,,,"
[STATS] report the number of times each server was selected One user reported that an indicator was missing in the statistics: the number of times each server was selected by load balancing. It is in fact the total number of sessions assigned to a server by the load balancing algorithm. It should directly reflect the weight for "fair" algorithms such as round-robin, since it will not account for persistant connections. It should help a lot tuning each server's weight depending on the load it receives.
2007-12-05 04:34:49 -05:00
/* sessions : current, max, limit, total */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%d,%d,%d,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* bytes : in, out */
"%lld,%lld,"
/* denied: req, resp */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* errors : request, connect, response */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,,,"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
",,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* server status : reflect frontend status */
"%s,"
/* rest of server: nothing */
[MEDIUM] add support for "maxqueue" to limit server queue overload This patch adds the "maxqueue" parameter to the server. This allows new sessions to be immediately rebalanced when the server's queue is filled. It's useful when session stickiness is just a performance boost (even a huge one) but not a requirement. This should only be used if session affinity isn't a hard functional requirement but provides performance boost by keeping server-local caches hot and compact). Absence of 'maxqueue' option means unlimited queue. When queue gets filled up to 'maxqueue' client session is moved from server-local queue to a global one.
2007-10-25 14:15:38 -04:00
",,,,,,,,"
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
/* pid, iid, sid, throttle, lbtot, tracked, type */
"%d,%d,0,,,,%d,"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* rate, rate_lim, rate_max */
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
"%u,%u,%u,"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* check_status, check_code, check_duration */
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
",,,",
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
px->id,
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->feconn, px->fe_counters.conn_max, px->maxconn, px->fe_counters.cum_sess,
px->fe_counters.bytes_in, px->fe_counters.bytes_out,
px->fe_counters.denied_req, px->fe_counters.denied_resp,
px->fe_counters.failed_req,
[CLEANUP] proxy: rename a few proxy states (PR_STIDLE and PR_STRUN) Those states have been replaced with PR_STFULL and PR_STREADY respectively, as it is what matches them the best now. Also, two occurrences of PR_STIDLE in peers.c have been removed as this did not provide any form of error recovery anyway.
2011-07-25 02:11:52 -04:00
px->state == PR_STREADY ? "OPEN" :
px->state == PR_STFULL ? "FULL" : "STOP",
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
relative_pid, px->uuid, STATS_TYPE_FE,
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
read_freq_ctr(&px->fe_sess_per_sec),
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->fe_sps_lim, px->fe_counters.sps_max);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
/* http response: 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
for (i=1; i<6; i++)
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, "%lld,", px->fe_counters.p.http.rsp[i]);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, "%lld,", px->fe_counters.p.http.rsp[0]);
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
} else {
chunk_printf(&msg, ",,,,,,");
}
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
/* failed health analyses */
chunk_printf(&msg, ",");
[STATS] report HTTP requests (total and rate) in frontends Now that we support keep-alive, it's important to report a separate counter for requests. Right now it just appears in the CSV output.
2010-02-26 04:05:55 -05:00
/* requests : req_rate, req_rate_max, req_tot, */
chunk_printf(&msg, "%u,%u,%lld,",
read_freq_ctr(&px->fe_req_per_sec),
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->fe_counters.p.http.rps_max, px->fe_counters.p.http.cum_req);
[STATS] report HTTP requests (total and rate) in frontends Now that we support keep-alive, it's important to report a separate counter for requests. Right now it just appears in the CSV output.
2010-02-26 04:05:55 -05:00
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors: cli_aborts, srv_aborts */
chunk_printf(&msg, ",,");
[MINOR] Collect & provide http response codes for frontends, fix backends This patch extends and corrects the functionality introduced by "Collect & provide http response codes received from servers": - responses are now also accounted for frontends - backend's and frontend's counters are incremented based on responses sent to client, not received from servers
2009-10-24 09:36:15 -04:00
/* finish with EOL */
chunk_printf(&msg, "\n");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 0;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.l = px->listen; /* may be NULL */
si->applet.ctx.stats.px_st = STAT_PX_ST_LI;
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_LI:
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* stats.l has been initialized above */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
for (; si->applet.ctx.stats.l != NULL; si->applet.ctx.stats.l = l->next) {
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
if (buffer_almost_full(&rep->buf))
[OPTIM] stats: check free space before trying to print This alone makes a typical HTML stats dump consume 10% CPU less, because we avoid doing complex printf calls to drop them later. Only a few common cases have been checked, those which are very likely to run for nothing.
2009-10-11 17:35:10 -04:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
l = si->applet.ctx.stats.l;
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
if (!l->counters)
continue;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_BOUND) {
if (!(si->applet.ctx.stats.type & (1 << STATS_TYPE_SO)))
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
break;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.sid != -1 && l->luid != si->applet.ctx.stats.sid)
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
continue;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg, "<tr class=socket>");
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* Column sub-heading for Enable or Disable server */
chunk_printf(&msg, "<td></td>");
}
chunk_printf(&msg, "<td class=ac");
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
if (uri->flags&ST_SHLGNDS) {
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
char str[INET6_ADDRSTRLEN];
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
int port;
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, " title=\"");
port = get_host_port(&l->addr);
switch (addr_to_str(&l->addr, str, sizeof(str))) {
case AF_INET:
chunk_printf(&msg, "IPv4: %s:%d, ", str, port);
break;
case AF_INET6:
chunk_printf(&msg, "IPv6: [%s]:%d, ", str, port);
break;
case AF_UNIX:
chunk_printf(&msg, "unix, ");
break;
case -1:
chunk_printf(&msg, "(%s), ", strerror(errno));
break;
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
}
/* id */
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, "id: %d\"", l->luid);
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
}
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
chunk_printf(&msg,
/* name, queue */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s<a name=\"%s/+%s\"></a>"
"<a class=lfsb href=\"#%s/+%s\">%s</a></td><td colspan=3>%s</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* sessions rate: current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td colspan=3>&nbsp;</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* sessions: current, max, limit, total, lbtot */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td>%s</td>"
"<td>%s</td><td>&nbsp;</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* bytes: in, out */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
"",
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"<u>":"",
[MINOR] stats: add "a link" & "a href" for sockets This patch adds add "a link" & "a href" html tags for sockets. As sockets may have the same name like servers, I decided to add "+" char (forbidden in names assigned to servers), as a prefix.
2010-01-04 05:33:32 -05:00
px->id, l->name, px->id, l->name, l->name,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"</u>":"",
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
U2H3(l->nbconn), U2H4(l->counters->conn_max), U2H5(l->maxconn),
U2H6(l->counters->cum_conn), U2H7(l->counters->bytes_in), U2H8(l->counters->bytes_out));
chunk_printf(&msg,
/* denied: req, resp */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* errors: request, connect, response */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td></td><td></td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* warnings: retries, redispatches */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td></td><td></td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* server status: reflect listener status */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>%s</td>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
/* rest of server: nothing */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac colspan=8></td></tr>"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
"",
U2H0(l->counters->denied_req), U2H1(l->counters->denied_resp),
U2H2(l->counters->failed_req),
[MINOR] stats: report a "WAITING" state for sockets waiting for resource This is useful when enabling socket-stats to know that a socket is being waiting for some resource (RAM, global connections, etc...).
2011-07-25 02:16:20 -04:00
(l->nbconn < l->maxconn) ? (l->state == LI_LIMITED) ? "WAITING" : "OPEN" : "FULL");
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
} else {
chunk_printf(&msg,
/* pxid, name, queue cur, queue max, */
"%s,%s,,,"
/* sessions: current, max, limit, total */
"%d,%d,%d,%lld,"
/* bytes: in, out */
"%lld,%lld,"
/* denied: req, resp */
"%lld,%lld,"
/* errors: request, connect, response */
"%lld,,,"
/* warnings: retries, redispatches */
",,"
/* server status: reflect listener status */
"%s,"
/* rest of server: nothing */
",,,,,,,,"
/* pid, iid, sid, throttle, lbtot, tracked, type */
"%d,%d,%d,,,,%d,"
/* rate, rate_lim, rate_max */
",,,"
/* check_status, check_code, check_duration */
",,,"
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* http response: 1xx, 2xx, 3xx, 4xx, 5xx, other */
",,,,,,"
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
/* failed health analyses */
","
[STATS] report HTTP requests (total and rate) in frontends Now that we support keep-alive, it's important to report a separate counter for requests. Right now it just appears in the CSV output.
2010-02-26 04:05:55 -05:00
/* requests : req_rate, req_rate_max, req_tot, */
",,,"
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors: cli_aborts, srv_aborts */
",,"
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
"\n",
px->id, l->name,
l->nbconn, l->counters->conn_max,
l->maxconn, l->counters->cum_conn,
l->counters->bytes_in, l->counters->bytes_out,
l->counters->denied_req, l->counters->denied_resp,
l->counters->failed_req,
(l->nbconn < l->maxconn) ? "OPEN" : "FULL",
relative_pid, px->uuid, l->luid, STATS_TYPE_SO);
}
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] Collect & provide separate statistics for sockets, v2 This patch allows to collect & provide separate statistics for each socket. It can be very useful if you would like to distinguish between traffic generate by local and remote users or between different types of remote clients (peerings, domestic, foreign). Currently no "Session rate" is supported, but adding it should be possible if we found it useful.
2009-10-04 09:43:17 -04:00
return 0;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.sv = px->srv; /* may be NULL */
si->applet.ctx.stats.px_st = STAT_PX_ST_SV;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_SV:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* stats.sv has been initialized above */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
for (; si->applet.ctx.stats.sv != NULL; si->applet.ctx.stats.sv = sv->next) {
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
int sv_state; /* 0=DOWN, 1=going up, 2=going down, 3=UP, 4,5=NOLB, 6=unchecked */
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
if (buffer_almost_full(&rep->buf))
[OPTIM] stats: check free space before trying to print This alone makes a typical HTML stats dump consume 10% CPU less, because we avoid doing complex printf calls to drop them later. Only a few common cases have been checked, those which are very likely to run for nothing.
2009-10-11 17:35:10 -04:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
sv = si->applet.ctx.stats.sv;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.flags & STAT_BOUND) {
if (!(si->applet.ctx.stats.type & (1 << STATS_TYPE_SV)))
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
break;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.stats.sid != -1 && sv->puid != si->applet.ctx.stats.sid)
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
continue;
}
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
if (sv->track)
svs = sv->track;
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
else
svs = sv;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* FIXME: produce some small strings for "UP/DOWN x/y &#xxxx;" */
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
if (!(svs->state & SRV_CHECKED))
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
sv_state = 6;
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
else if (svs->state & SRV_RUNNING) {
if (svs->health == svs->rise + svs->fall - 1)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
sv_state = 3; /* UP */
else
sv_state = 2; /* going down */
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
if (svs->state & SRV_GOINGDOWN)
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
sv_state += 2;
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
else
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
if (svs->health)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
sv_state = 1; /* going up */
else
sv_state = 0; /* DOWN */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (((sv_state == 0) || (sv->state & SRV_MAINTAIN)) && (si->applet.ctx.stats.flags & STAT_HIDE_DOWN)) {
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* do not report servers which are DOWN */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.sv = sv->next;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
continue;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
static char *srv_hlt_st[7] = { "DOWN", "DN %d/%d &uarr;",
"UP %d/%d &darr;", "UP",
"NOLB %d/%d &darr;", "NOLB",
"<i>no check</i>" };
[MINOR] stats: report maint state for tracking servers too This is to avoid confusing a real server failure and a tracking server which is marked down because the tracked server was switched to MAINT.
2010-02-02 18:26:28 -05:00
if ((sv->state & SRV_MAINTAIN) || (svs->state & SRV_MAINTAIN)) {
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
chunk_printf(&msg,
/* name */
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"<tr class=\"maintain\">"
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
);
}
else {
chunk_printf(&msg,
/* name */
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"<tr class=\"%s%d\">",
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
(sv->state & SRV_BACKUP) ? "backup" : "active", sv_state);
}
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] stats: add an admin level The stats web interface must be read-only by default to prevent security holes. As it is now allowed to enable/disable servers, a new keyword "stats admin" is introduced to activate this admin level, conditioned by ACLs. (cherry picked from commit 5334bab92ca7debe36df69983c19c21b6dc63f78)
2010-10-11 18:14:36 -04:00
chunk_printf(&msg,
"<td><input type=\"checkbox\" name=\"s\" value=\"%s\"></td>",
sv->id);
}
chunk_printf(&msg, "<td class=ac");
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
if (uri->flags&ST_SHLGNDS) {
char str[INET6_ADDRSTRLEN];
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, " title=\"");
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
switch (addr_to_str(&sv->addr, str, sizeof(str))) {
[MEDIUM] add internal support for IPv6 server addresses This patch turns internal server addresses to sockaddr_storage to store IPv6 addresses, and makes the connect() function use it. This code already works but some caveats with getaddrinfo/gethostbyname still need to be sorted out while the changes had to be merged at this stage of internal architecture changes. So for now the config parser will not emit an IPv6 address yet so that user experience remains unchanged. This change should have absolutely zero user-visible effect, otherwise it's a bug introduced during the merge, that should be reported ASAP.
2011-03-10 16:26:24 -05:00
case AF_INET:
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, "IPv4: %s:%d, ", str, get_host_port(&sv->addr));
[MEDIUM] add internal support for IPv6 server addresses This patch turns internal server addresses to sockaddr_storage to store IPv6 addresses, and makes the connect() function use it. This code already works but some caveats with getaddrinfo/gethostbyname still need to be sorted out while the changes had to be merged at this stage of internal architecture changes. So for now the config parser will not emit an IPv6 address yet so that user experience remains unchanged. This change should have absolutely zero user-visible effect, otherwise it's a bug introduced during the merge, that should be reported ASAP.
2011-03-10 16:26:24 -05:00
break;
case AF_INET6:
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, "IPv6: [%s]:%d, ", str, get_host_port(&sv->addr));
break;
case AF_UNIX:
chunk_printf(&msg, "unix, ");
break;
case -1:
chunk_printf(&msg, "(%s), ", strerror(errno));
break;
default: /* address family not supported */
[MEDIUM] add internal support for IPv6 server addresses This patch turns internal server addresses to sockaddr_storage to store IPv6 addresses, and makes the connect() function use it. This code already works but some caveats with getaddrinfo/gethostbyname still need to be sorted out while the changes had to be merged at this stage of internal architecture changes. So for now the config parser will not emit an IPv6 address yet so that user experience remains unchanged. This change should have absolutely zero user-visible effect, otherwise it's a bug introduced during the merge, that should be reported ASAP.
2011-03-10 16:26:24 -05:00
break;
}
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
/* id */
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, "id: %d", sv->puid);
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
/* cookie */
if (sv->cookie) {
struct chunk src;
chunk_printf(&msg, ", cookie: '");
chunk_initlen(&src, sv->cookie, 0, strlen(sv->cookie));
chunk_htmlencode(&msg, &src);
chunk_printf(&msg, "'");
}
chunk_printf(&msg, "\"");
}
chunk_printf(&msg,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s<a name=\"%s/%s\"></a>"
"<a class=lfsb href=\"#%s/%s\">%s</a>%s</td>"
[MEDIUM] add support for "maxqueue" to limit server queue overload This patch adds the "maxqueue" parameter to the server. This allows new sessions to be immediately rebalanced when the server's queue is filled. It's useful when session stickiness is just a performance boost (even a huge one) but not a requirement. This should only be used if session affinity isn't a hard functional requirement but provides performance boost by keeping server-local caches hot and compact). Absence of 'maxqueue' option means unlimited queue. When queue gets filled up to 'maxqueue' client session is moved from server-local queue to a global one.
2007-10-25 14:15:38 -04:00
/* queue : current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td>%s</td>"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
/* sessions rate : current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td></td>"
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* sessions: current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td>%s</td>"
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
"<td"
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
"",
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"<u>":"",
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
px->id, sv->id, px->id, sv->id, sv->id,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"</u>":"",
[OPTIM] counters: move some max numbers to the counters struct There are a few remaining max values that need to move to counters. Also, the counters are more often used than some config information, so get them closer to the other useful struct members for better cache efficiency.
2009-10-04 17:12:44 -04:00
U2H0(sv->nbpend), U2H1(sv->counters.nbpend_max), LIM2A2(sv->maxqueue, "-"),
U2H3(read_freq_ctr(&sv->sess_per_sec)), U2H4(sv->counters.sps_max),
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
U2H5(sv->cur_sess), U2H6(sv->counters.cur_sess_max), LIM2A7(sv->maxconn, "-"));
/* http response (via td title): 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
chunk_printf(&msg, " title=\"rsp codes:");
for (i = 1; i < 6; i++)
chunk_printf(&msg, " %dxx=%lld,", i, sv->counters.p.http.rsp[i]);
chunk_printf(&msg, " other=%lld\"", sv->counters.p.http.rsp[0]);
}
chunk_printf(&msg,
/* sessions: total, lbtot */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s%s%s</td><td>%s</td>",
(px->mode == PR_MODE_HTTP)?"<u>":"",
U2H0(sv->counters.cum_sess),
(px->mode == PR_MODE_HTTP)?"</u>":"",
U2H1(sv->counters.cum_lbconn));
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* bytes : in, out */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* denied: req, resp */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td></td><td>%s</td>"
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors : request, connect */
"<td></td><td>%s</td>"
/* errors : response */
[BUG] stats: connection reset counters must be plain ascii, not HTML Using U2H*() to put numbers in title tags is a bad idea as those appear in the HTML itself. Problem reported by Laurent Dolosor.
2010-03-05 12:15:23 -05:00
"<td title=\"Connection resets during transfers: %lld client, %lld server\"><u>%s</u></td>"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%lld</td><td>%lld</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"",
[CLEANUP] Move counters to dedicated structures Move counters from "struct proxy" and "struct server" to "struct pxcounters" and "struct svcounters". This patch should make no functional change.
2009-10-04 08:52:57 -04:00
U2H0(sv->counters.bytes_in), U2H1(sv->counters.bytes_out),
U2H2(sv->counters.failed_secu),
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
U2H3(sv->counters.failed_conns),
[BUG] stats: connection reset counters must be plain ascii, not HTML Using U2H*() to put numbers in title tags is a bad idea as those appear in the HTML itself. Problem reported by Laurent Dolosor.
2010-03-05 12:15:23 -05:00
sv->counters.cli_aborts,
sv->counters.srv_aborts,
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
U2H6(sv->counters.failed_resp),
[CLEANUP] Move counters to dedicated structures Move counters from "struct proxy" and "struct server" to "struct pxcounters" and "struct svcounters". This patch should make no functional change.
2009-10-04 08:52:57 -04:00
sv->counters.retries, sv->counters.redispatches);
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* status, lest check */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
chunk_printf(&msg, "<td class=ac>");
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
if (sv->state & SRV_MAINTAIN) {
chunk_printf(&msg, "%s ",
human_time(now.tv_sec - sv->last_change, 1));
chunk_printf(&msg, "MAINT");
}
[MINOR] stats: report maint state for tracking servers too This is to avoid confusing a real server failure and a tracking server which is marked down because the tracked server was switched to MAINT.
2010-02-02 18:26:28 -05:00
else if (svs != sv && svs->state & SRV_MAINTAIN) {
chunk_printf(&msg, "%s ",
human_time(now.tv_sec - svs->last_change, 1));
chunk_printf(&msg, "MAINT(via)");
}
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
else if (svs->state & SRV_CHECKED) {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%s ",
[MINOR] stats: report maint state for tracking servers too This is to avoid confusing a real server failure and a tracking server which is marked down because the tracked server was switched to MAINT.
2010-02-02 18:26:28 -05:00
human_time(now.tv_sec - svs->last_change, 1));
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
srv_hlt_st[sv_state],
(svs->state & SRV_RUNNING) ? (svs->health - svs->rise + 1) : (svs->health),
(svs->state & SRV_RUNNING) ? (svs->fall) : (svs->rise));
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
}
[MINOR] Capture & display more data from health checks, v2 Capture & display more data from health checks, like strerror(errno) for L4 failed checks or a first line from a response for L7 successes/failed checks. Non ascii or control characters are masked with chunk_htmlencode() (html stats) or chunk_asciiencode() (logs).
2009-10-10 15:06:49 -04:00
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
if (sv->state & SRV_CHECKED) {
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
chunk_printf(&msg, "</td><td class=ac title=\"%s",
[MINOR] Capture & display more data from health checks, v2 Capture & display more data from health checks, like strerror(errno) for L4 failed checks or a first line from a response for L7 successes/failed checks. Non ascii or control characters are masked with chunk_htmlencode() (html stats) or chunk_asciiencode() (logs).
2009-10-10 15:06:49 -04:00
get_check_status_description(sv->check_status));
if (*sv->check_desc) {
struct chunk src;
chunk_printf(&msg, ": ");
chunk_initlen(&src, sv->check_desc, 0, strlen(sv->check_desc));
chunk_htmlencode(&msg, &src);
}
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
chunk_printf(&msg, "\"><u> %s%s",
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
tv_iszero(&sv->check_start)?"":"* ",
get_check_status_info(sv->check_status));
if (sv->check_status >= HCHK_STATUS_L57DATA)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "/%d", sv->check_code);
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
if (sv->check_status >= HCHK_STATUS_CHECKED && sv->check_duration >= 0)
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
chunk_printf(&msg, " in %lums</u>", sv->check_duration);
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
} else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "</td><td>");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* weight */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"</td><td class=ac>%d</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* act, bck */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>%s</td><td class=ac>%s</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
"",
[MEDIUM] slowstart: ensure we don't start with a null weight Because of a divide, it was possible to have a null weight during a slowstart, which is pretty annoying, especially with a single server and a long slowstart. Also, fix the way we report the values in the stats page to avoid confusion.
2007-12-02 20:04:00 -05:00
(sv->eweight * px->lbprm.wmult + px->lbprm.wdiv - 1) / px->lbprm.wdiv,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
(sv->state & SRV_BACKUP) ? "-" : "Y",
(sv->state & SRV_BACKUP) ? "Y" : "-");
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
/* check failures: unique, fatal, down time */
[MINOR] hana: only report stats if it is enabled It is useless to report statistics if the feature was not enabled. It also makes possible to distinguish if health analyses is enabled or not only by looking at the stats page.
2010-01-03 18:48:43 -05:00
if (sv->state & SRV_CHECKED) {
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
chunk_printf(&msg, "<td title=\"Failed Health Checks%s\"><u>%lld",
[MINOR] hana: only report stats if it is enabled It is useless to report statistics if the feature was not enabled. It also makes possible to distinguish if health analyses is enabled or not only by looking at the stats page.
2010-01-03 18:48:43 -05:00
svs->observe?"/Health Analyses":"", svs->counters.failed_checks);
if (svs->observe)
chunk_printf(&msg, "/%lld", svs->counters.failed_hana);
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
"</u></td>"
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
"<td>%lld</td><td>%s</td>"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"",
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
svs->counters.down_trans, human_time(srv_downtime(sv), 1));
[MINOR] hana: only report stats if it is enabled It is useless to report statistics if the feature was not enabled. It also makes possible to distinguish if health analyses is enabled or not only by looking at the stats page.
2010-01-03 18:48:43 -05:00
} else if (sv != svs)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] stats: add href to tracked server Each server has now a dedicated "a link" tag, so let's use it.
2010-01-05 12:33:01 -05:00
"<td class=ac colspan=3><a class=lfsb href=\"#%s/%s\">via %s/%s<a></td>",
svs->proxy->id, svs->id, svs->proxy->id, svs->id);
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"<td colspan=3></td>");
/* throttle */
if ((sv->state & SRV_WARMINGUP) &&
now.tv_sec < sv->last_change + sv->slowstart &&
now.tv_sec >= sv->last_change) {
unsigned int ratio;
ratio = MAX(1, 100 * (now.tv_sec - sv->last_change) / sv->slowstart);
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>%d %%</td></tr>\n", ratio);
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
} else {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>-</td></tr>\n");
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
}
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
} else {
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
static char *srv_hlt_st[7] = { "DOWN,", "DOWN %d/%d,",
"UP %d/%d,", "UP,",
"NOLB %d/%d,", "NOLB,",
"no check," };
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* pxid, name */
"%s,%s,"
/* queue : current, max */
"%d,%d,"
[STATS] report the number of times each server was selected One user reported that an indicator was missing in the statistics: the number of times each server was selected by load balancing. It is in fact the total number of sessions assigned to a server by the load balancing algorithm. It should directly reflect the weight for "fair" algorithms such as round-robin, since it will not account for persistant connections. It should help a lot tuning each server's weight depending on the load it receives.
2007-12-05 04:34:49 -05:00
/* sessions : current, max, limit, total */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%d,%d,%s,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* bytes : in, out */
"%lld,%lld,"
/* denied: req, resp */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
",%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* errors : request, connect, response */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
",%lld,%lld,"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
"",
px->id, sv->id,
[OPTIM] counters: move some max numbers to the counters struct There are a few remaining max values that need to move to counters. Also, the counters are more often used than some config information, so get them closer to the other useful struct members for better cache efficiency.
2009-10-04 17:12:44 -04:00
sv->nbpend, sv->counters.nbpend_max,
sv->cur_sess, sv->counters.cur_sess_max, LIM2A0(sv->maxconn, ""), sv->counters.cum_sess,
[CLEANUP] Move counters to dedicated structures Move counters from "struct proxy" and "struct server" to "struct pxcounters" and "struct svcounters". This patch should make no functional change.
2009-10-04 08:52:57 -04:00
sv->counters.bytes_in, sv->counters.bytes_out,
sv->counters.failed_secu,
sv->counters.failed_conns, sv->counters.failed_resp,
sv->counters.retries, sv->counters.redispatches);
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* status */
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
if (sv->state & SRV_MAINTAIN) {
chunk_printf(&msg, "MAINT,");
[MINOR] stats: report maint state for tracking servers too This is to avoid confusing a real server failure and a tracking server which is marked down because the tracked server was switched to MAINT.
2010-02-02 18:26:28 -05:00
}
else if (svs != sv && svs->state & SRV_MAINTAIN) {
chunk_printf(&msg, "MAINT(via),");
}
else {
[MEDIUM] add a maintenance mode to servers This is a first attempt to add a maintenance mode on servers, using the stat socket (in admin level). It can be done with the following command : - disable server <backend>/<server> - enable server <backend>/<server> In this mode, no more checks will be performed on the server and it will be marked as a special DOWN state (MAINT). If some servers were tracking it, they'll go DOWN until the server leaves the maintenance mode. The stats page and the CSV export also display this special state. This can be used to disable the server in haproxy before doing some operations on this server itself. This is a good complement to the "http-check disable-on-404" keyword and works in TCP mode.
2010-01-31 16:34:03 -05:00
chunk_printf(&msg,
srv_hlt_st[sv_state],
(svs->state & SRV_RUNNING) ? (svs->health - svs->rise + 1) : (svs->health),
(svs->state & SRV_RUNNING) ? (svs->fall) : (svs->rise));
}
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* weight, active, backup */
"%d,%d,%d,"
"",
[MEDIUM] slowstart: ensure we don't start with a null weight Because of a divide, it was possible to have a null weight during a slowstart, which is pretty annoying, especially with a single server and a long slowstart. Also, fix the way we report the values in the stats page to avoid confusion.
2007-12-02 20:04:00 -05:00
(sv->eweight * px->lbprm.wmult + px->lbprm.wdiv - 1) / px->lbprm.wdiv,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
(sv->state & SRV_BACKUP) ? 0 : 1,
(sv->state & SRV_BACKUP) ? 1 : 0);
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
/* check failures: unique, fatal; last change, total downtime */
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
if (sv->state & SRV_CHECKED)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,%lld,%d,%d,",
[CLEANUP] Move counters to dedicated structures Move counters from "struct proxy" and "struct server" to "struct pxcounters" and "struct svcounters". This patch should make no functional change.
2009-10-04 08:52:57 -04:00
sv->counters.failed_checks, sv->counters.down_trans,
[MINOR] fix several printf formats and missing arguments Last patch revealed a number of mistakes in printf-like calls, mostly int/long mismatches, and a few missing arguments.
2009-04-03 08:49:12 -04:00
(int)(now.tv_sec - sv->last_change), srv_downtime(sv));
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] add support for "maxqueue" to limit server queue overload This patch adds the "maxqueue" parameter to the server. This allows new sessions to be immediately rebalanced when the server's queue is filled. It's useful when session stickiness is just a performance boost (even a huge one) but not a requirement. This should only be used if session affinity isn't a hard functional requirement but provides performance boost by keeping server-local caches hot and compact). Absence of 'maxqueue' option means unlimited queue. When queue gets filled up to 'maxqueue' client session is moved from server-local queue to a global one.
2007-10-25 14:15:38 -04:00
",,,,");
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
/* queue limit, pid, iid, sid, */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] stats: report numerical process ID, proxy ID and server ID It is very convenient for SNMP monitoring to have unique process ID, proxy ID and server ID. Those have been added to the CSV outputs. The numbers start at 1. 0 is reserved. For servers, 0 means that the reported name is not a server name but half a proxy (FRONTEND/BACKEND). A remaining hidden "-" in the CSV output has been eliminated too.
2007-11-04 17:35:08 -05:00
"%s,"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"%d,%d,%d,",
[MINOR] stats: report numerical process ID, proxy ID and server ID It is very convenient for SNMP monitoring to have unique process ID, proxy ID and server ID. Those have been added to the CSV outputs. The numbers start at 1. 0 is reserved. For servers, 0 means that the reported name is not a server name but half a proxy (FRONTEND/BACKEND). A remaining hidden "-" in the CSV output has been eliminated too.
2007-11-04 17:35:08 -05:00
LIM2A0(sv->maxqueue, ""),
relative_pid, px->uuid, sv->puid);
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
/* throttle */
if ((sv->state & SRV_WARMINGUP) &&
now.tv_sec < sv->last_change + sv->slowstart &&
now.tv_sec >= sv->last_change) {
unsigned int ratio;
ratio = MAX(1, 100 * (now.tv_sec - sv->last_change) / sv->slowstart);
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%d", ratio);
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
}
[STATS] report the number of times each server was selected One user reported that an indicator was missing in the statistics: the number of times each server was selected by load balancing. It is in fact the total number of sessions assigned to a server by the load balancing algorithm. It should directly reflect the weight for "fair" algorithms such as round-robin, since it will not account for persistant connections. It should help a lot tuning each server's weight depending on the load it receives.
2007-12-05 04:34:49 -05:00
/* sessions: lbtot */
[CLEANUP] Move counters to dedicated structures Move counters from "struct proxy" and "struct server" to "struct pxcounters" and "struct svcounters". This patch should make no functional change.
2009-10-04 08:52:57 -04:00
chunk_printf(&msg, ",%lld,", sv->counters.cum_lbconn);
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
/* tracked */
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
if (sv->track)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%s/%s,",
CLEANUP: rename possibly confusing struct field "tracked" When reading the code, the "tracked" member of a server makes one think the server is tracked while it's the opposite, it's a pointer to the server being tracked. This is particularly true in constructs such as : if (srv->tracked) { Since it's the second time I get caught misunderstanding it, let's rename it to "track" to avoid the confusion.
2011-10-28 09:35:33 -04:00
sv->track->proxy->id, sv->track->id);
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, ",");
[MEDIUM] Implement "track [<backend>/]<server>" This patch implements ability to set the current state of one server by tracking another one. It: - adds two variables: *tracknext, *tracked to struct server - implements findserver(), similar to findproxy() - adds "track" keyword accepting both "proxy/server" and "server" (assuming current proxy) - verifies if both checks and tracking is not enabled at the same time - changes set_server_down() to notify tracking server - creates set_server_up(), set_server_disabled(), set_server_enabled() by moving the code from process_chk() and adding notifications - changes stats to show a name of tracked server instead of Chk/Dwn/Dwntime(html) or by adding new variable (csv) Changes from the previuos version: - it is possibile to track independently of the declaration order - one extra comma bug is fixed - new condition to check if there is no disable-on-404 inconsistency
2008-02-17 19:26:35 -05:00
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
/* type */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%d,", STATS_TYPE_SV);
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
/* rate */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%u,,%u,",
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
read_freq_ctr(&sv->sess_per_sec),
[OPTIM] counters: move some max numbers to the counters struct There are a few remaining max values that need to move to counters. Also, the counters are more often used than some config information, so get them closer to the other useful struct members for better cache efficiency.
2009-10-04 17:12:44 -04:00
sv->counters.sps_max);
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
if (sv->state & SRV_CHECKED) {
/* check_status */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%s,", get_check_status_info(sv->check_status));
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* check_code */
if (sv->check_status >= HCHK_STATUS_L57DATA)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%u,", sv->check_code);
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, ",");
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* check_duration */
[BUG] stats: show UP/DOWN status also in tracking servers It was broken since 0960541e496167f4cc5f89f114e027bf8f7f1ff0: [MEDIUM] Collect & show information about last health check, v3
2010-01-05 12:44:44 -05:00
if (sv->check_status >= HCHK_STATUS_CHECKED)
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "%lu,", sv->check_duration);
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
else
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, ",");
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
} else {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, ",,,");
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
}
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* http response: 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
for (i=1; i<6; i++)
chunk_printf(&msg, "%lld,", sv->counters.p.http.rsp[i]);
chunk_printf(&msg, "%lld,", sv->counters.p.http.rsp[0]);
} else {
chunk_printf(&msg, ",,,,,,");
}
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
/* failed health analyses */
chunk_printf(&msg, "%lld,", sv->counters.failed_hana);
[STATS] report HTTP requests (total and rate) in frontends Now that we support keep-alive, it's important to report a separate counter for requests. Right now it just appears in the CSV output.
2010-02-26 04:05:55 -05:00
/* requests : req_rate, req_rate_max, req_tot, */
chunk_printf(&msg, ",,,");
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors: cli_aborts, srv_aborts */
chunk_printf(&msg, "%lld,%lld,",
sv->counters.cli_aborts, sv->counters.srv_aborts);
[MEDIUM] measure and report session rate on frontend, backends and servers With this change, all frontends, backends, and servers maintain a session counter and a timer to compute a session rate over the last second. This value will be very useful because it varies instantly and can be used to check thresholds. This value is also reported in the stats in a new "rate" column.
2009-03-05 12:43:00 -05:00
/* finish with EOL */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "\n");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
}
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 0;
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
} /* for sv */
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px_st = STAT_PX_ST_BE;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_BE:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* print the backend */
[MEDIUM] fix stats socket limitation to 16 kB Due to the way the stats socket work, it was not possible to maintain the information related to the command entered, so after filling a whole buffer, the request was lost and it was considered that there was nothing to write anymore. The major reason was that some flags were passed directly during the first call to stats_dump_raw() instead of being stored persistently in the session. To definitely fix this problem, flags were added to the stats member of the session structure. A second problem appeared. When the stats were produced, a first call to client_retnclose() was performed, then one or multiple subsequent calls to buffer_write_chunks() were done. But once the stats buffer was full and a reschedule operated, the buffer was flushed, the write flag cleared from the buffer and nothing was done to re-arm it. For this reason, a check was added in the proto_uxst_stats() function in order to re-call the client FSM when data were added by stats_dump_raw(). Finally, the whole unix stats dump FSM was rewritten to avoid all the magics it depended on. It is now simpler and looks more like the HTTP one.
2008-03-17 16:38:24 -04:00
if ((px->cap & PR_CAP_BE) &&
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
(!(si->applet.ctx.stats.flags & STAT_BOUND) || (si->applet.ctx.stats.type & (1 << STATS_TYPE_BE)))) {
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg, "<tr class=\"backend\">");
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* Column sub-heading for Enable or Disable server */
chunk_printf(&msg, "<td></td>");
}
chunk_printf(&msg, "<td class=ac");
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
if (uri->flags&ST_SHLGNDS) {
/* balancing */
[BUG] stats: cookie should be reported under backend not under proxy
2010-01-06 09:03:18 -05:00
chunk_printf(&msg, " title=\"balancing: %s",
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
backend_lb_algo_str(px->lbprm.algo & BE_LB_ALGO));
[BUG] stats: cookie should be reported under backend not under proxy
2010-01-06 09:03:18 -05:00
/* cookie */
if (px->cookie_name) {
struct chunk src;
chunk_printf(&msg, ", cookie: '");
chunk_initlen(&src, px->cookie_name, 0, strlen(px->cookie_name));
chunk_htmlencode(&msg, &src);
chunk_printf(&msg, "'");
}
chunk_printf(&msg, "\"");
[MINOR]: stats: add show-legends to report additional informations Supported informations, available via "tr/td title": - cap: capabilities (proxy) - mode: one of tcp, http or health (proxy) - id: SNMP ID (proxy, socket, server) - IP (socket, server) - cookie (backend, server)
2010-01-04 10:03:09 -05:00
}
chunk_printf(&msg,
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* name */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s<a name=\"%s/Backend\"></a>"
"<a class=lfsb href=\"#%s/Backend\">Backend</a>%s</td>"
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* queue : current, max */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td></td>"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
/* sessions rate : current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td></td>"
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
"",
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"<u>":"",
[MINOR] add additional "a href"s to stats page This patch adds <a href> html links for proxies, frontends, servers and backends. Once located, can be clicked. Users no longer have to manually add #anchor to stat's url.
2009-10-24 08:24:30 -04:00
px->id, px->id,
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(uri->flags & ST_SHLGNDS)?"</u>":"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H0(px->nbpend) /* or px->totpend ? */, U2H1(px->be_counters.nbpend_max),
U2H2(read_freq_ctr(&px->be_sess_per_sec)), U2H3(px->be_counters.sps_max));
[MINOR] stats: report max sessions/s and limit in HTML stats
2009-05-10 13:19:41 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* sessions: current, max, limit */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td><td>%s</td>"
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
"<td"
"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H2(px->beconn), U2H3(px->be_counters.conn_max), U2H4(px->fullconn));
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* http response (via td title): 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
chunk_printf(&msg, " title=\"rsp codes:");
for (i = 1; i < 6; i++)
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, " %dxx=%lld", i, px->be_counters.p.http.rsp[i]);
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, " other=%lld\"", px->be_counters.p.http.rsp[0]);
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
}
chunk_printf(&msg,
/* sessions: total, lbtot */
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
">%s%s%s</td><td>%s</td>"
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* bytes: in, out */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
"",
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(px->mode == PR_MODE_HTTP)?"<u>":"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H6(px->be_counters.cum_conn),
[MINOR] stats: mark areas with an underline when tooltips are available There are many information available in the stats page that can only be seen when the mouse hovers over them. But it's hard to know where those information are. Now with a discrete dotted underline it's easier to spot those areas.
2010-02-26 06:29:07 -05:00
(px->mode == PR_MODE_HTTP)?"</u>":"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H7(px->be_counters.cum_lbconn),
U2H8(px->be_counters.bytes_in), U2H9(px->be_counters.bytes_out));
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* denied: req, resp */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%s</td><td>%s</td>"
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors : request, connect */
"<td></td><td>%s</td>"
/* errors : response */
[BUG] stats: connection reset counters must be plain ascii, not HTML Using U2H*() to put numbers in title tags is a bad idea as those appear in the HTML itself. Problem reported by Laurent Dolosor.
2010-03-05 12:15:23 -05:00
"<td title=\"Connection resets during transfers: %lld client, %lld server\"><u>%s</u></td>"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td>%lld</td><td>%lld</td>"
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
/* backend status: reflect backend status (up/down): we display UP
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
* if the backend has known working servers or if it has no server at
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
* all (eg: for stats). Then we display the total weight, number of
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
* active and backups. */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>%s %s</td><td class=ac>&nbsp;</td><td class=ac>%d</td>"
"<td class=ac>%d</td><td class=ac>%d</td>"
[MINOR] stats/html: group digits by 3 to clarify numbers Large stats numbers are more readable if there is a small space between groups of 3 digits.
2009-03-29 08:46:01 -04:00
"",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
U2H0(px->be_counters.denied_req), U2H1(px->be_counters.denied_resp),
U2H2(px->be_counters.failed_conns),
px->be_counters.cli_aborts,
px->be_counters.srv_aborts,
U2H5(px->be_counters.failed_resp),
px->be_counters.retries, px->be_counters.redispatches,
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
human_time(now.tv_sec - px->last_change, 1),
[MEDIUM] report disabled servers as "NOLB" when they are still UP It's important to be able to distinguish between servers which are UP and those which are UP but disabled via a 404 response. For this reason, the status entries report "NOLB" instead of "UP", and the HTML page uses darker colors. As a complement, write "DOWN" in bold red on the backend if it has no server left for load balancing.
2007-11-30 06:04:38 -05:00
(px->lbprm.tot_weight > 0 || !px->srv) ? "UP" :
"<font color=\"red\"><b>DOWN</b></font>",
[MEDIUM] slowstart: ensure we don't start with a null weight Because of a divide, it was possible to have a null weight during a slowstart, which is pretty annoying, especially with a single server and a long slowstart. Also, fix the way we report the values in the stats page to avoid confusion.
2007-12-02 20:04:00 -05:00
(px->lbprm.tot_weight * px->lbprm.wmult + px->lbprm.wdiv - 1) / px->lbprm.wdiv,
[MEDIUM] differentiate between generic LB params and map-specific ones Since the introduction of server weights, all load balancing algorithms relied on a pre-computed map. Incidently, quite a bunch of map-specific parameters were used at random places in order to get the number of servers or their total weight. It was not architecturally acceptable that optimizations for the map computation had impact on external parts. For instance, during this cleanup it was found that a backend weight was seen as 1 when only the first backup server is used, whatever its weight. This cleanup consists in differentiating between LB-generic parameters, such as total weights, number of servers, etc... and map-specific ones. The struct proxy has been enhanced in order to make it easier to later support other algorithms. The recount_servers() function now also updates generic values such as total weights so that it's not needed anymore to call recalc_server_map() when weights are needed. This permitted to simplify some code which does not need to know about map internals anymore.
2007-11-15 17:26:18 -05:00
px->srv_act, px->srv_bck);
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
/* rest of backend: nothing, down transitions, total downtime, throttle */
[MINOR] CSS & HTML fun This patch makes stats page about 30% smaller and "CSS 2.1" + "HTML 4.01 Transitional" compliant. There should be no visible differences. Changes: - add DOCTYPE for HTML 4.01 Transitional - add missing </ul> - remove cols=, AFAIK no modern browser support this property and it prevents validation to pass. - remove "align: center": there is no such property in css. There is however "text-align: center" but it is definitely not what we would like to see here. - by default align .titre to center - by default align .td to right - remove all align=right, no longer necessary - add class=ac (align center): shorter than "align=center" and use it when necessary - remove nowrap from td, instead use "white-space: nowrap" in css Now stats page passes W3C validators for HTML & CSS. We may consider adding "validated" icons from www.w3.org. ;)
2009-10-12 17:09:08 -04:00
"<td class=ac>&nbsp;</td><td>%d</td>"
"<td>%s</td>"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
"<td></td>"
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
"</tr>",
px->down_trans,
px->srv?human_time(be_downtime(px), 1):"&nbsp;");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
} else {
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* pxid, name */
"%s,BACKEND,"
/* queue : current, max */
"%d,%d,"
[STATS] report the number of times each server was selected One user reported that an indicator was missing in the statistics: the number of times each server was selected by load balancing. It is in fact the total number of sessions assigned to a server by the load balancing algorithm. It should directly reflect the weight for "fair" algorithms such as round-robin, since it will not account for persistant connections. It should help a lot tuning each server's weight depending on the load it receives.
2007-12-05 04:34:49 -05:00
/* sessions : current, max, limit, total */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%d,%d,%d,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* bytes : in, out */
"%lld,%lld,"
/* denied: req, resp */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,%lld,"
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
/* errors : request, connect, response */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
",%lld,%lld,"
[MEDIUM] stats: count server retries and redispatches It is important to know how your installation performs. Haproxy masks connection errors, which is extremely good for a client but it is bad for an administrator (except people believing that "ignorance is a bless"). Attached patch adds retries and redispatches counters, so now haproxy: 1. For server: - counts retried connections (masked or not) 2. For backends: - counts retried connections (masked or not) that happened to a slave server - counts redispatched connections - does not count successfully redispatched connections as backend errors. Errors are increased only when client does not get a valid response, in other words: with failed redispatch or when this function is not enabled. 3. For statistics: - display Retr (retries) and Redis (redispatches) as a "Warning" information.
2007-10-18 13:12:30 -04:00
/* warnings: retries, redispatches */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%lld,%lld,"
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
/* backend status: reflect backend status (up/down): we display UP
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
* if the backend has known working servers or if it has no server at
[MEDIUM] stats: report server and backend cumulated downtime Hello, This patch implements new statistics for SLA calculation by adding new field 'Dwntime' with total down time since restart (both HTTP/CSV) and extending status field (HTTP) or inserting a new one (CSV) with time showing how long each server/backend is in a current state. Additionaly, down transations are also calculated and displayed for backends, so it is possible to know how many times selected backend was down, generating "No server is available to handle this request." error. New information are presentetd in two different ways: - for HTTP: a "human redable form", one of "100000d 23h", "23h 59m" or "59m 59s" - for CSV: seconds I believe that seconds resolution is enough. As there are more columns in the status page I decided to shrink some names to make more space: - Weight -> Wght - Check -> Chk - Down -> Dwn Making described changes I also made some improvements and fixed some small bugs: - don't increment s->health above 's->rise + s->fall - 1'. Previously it was incremented an then (re)set to 's->rise + s->fall - 1'. - do not set server down if it is down already - do not set server up if it is up already - fix colspan in multiple places (mostly introduced by my previous patch) - add missing "status" header to CSV - fix order of retries/redispatches in server (CSV) - s/Tthen/Then/ - s/server/backend/ in DATA_ST_PX_BE (dumpstats.c) Changes from previous version: - deal with negative time intervales - don't relay on s->state (SRV_RUNNING) - little reworked human_time + compacted format (no spaces). If needed it can be used in the future for other purposes by optionally making "cnt" as an argument - leave set_server_down mostly unchanged - only little reworked "process_chk: 9" - additional fields in CSV are appended to the rigth - fix "SEC" macro - named arguments (human_time, be_downtime, srv_downtime) Hope it is OK. If there are only cosmetic changes needed please fill free to correct it, however if there are some bigger changes required I would like to discuss it first or at last to know what exactly was changed especially since I already put this patch into my production server. :) Thank you, Best regards, Krzysztof Oledzki
2007-10-22 10:21:10 -04:00
* all (eg: for stats). Then we display the total weight, number of
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
* active and backups. */
"%s,"
"%d,%d,%d,"
[MINOR] stats: report the server warm up status in a "throttle" column A new "throttle" column has been added to HTML and RAW stats to indicate in percent, the level of throttling due to server warmup. The column is empty at 100%.
2007-11-30 12:16:29 -05:00
/* rest of backend: nothing, down transitions, last change, total downtime */
[MEDIUM] add support for "maxqueue" to limit server queue overload This patch adds the "maxqueue" parameter to the server. This allows new sessions to be immediately rebalanced when the server's queue is filled. It's useful when session stickiness is just a performance boost (even a huge one) but not a requirement. This should only be used if session affinity isn't a hard functional requirement but provides performance boost by keeping server-local caches hot and compact). Absence of 'maxqueue' option means unlimited queue. When queue gets filled up to 'maxqueue' client session is moved from server-local queue to a global one.
2007-10-25 14:15:38 -04:00
",%d,%d,%d,,"
[MAJOR] proto_uxst rework -> SNMP support Currently there is a ~16KB limit for a data size passed via unix socket. It is caused by a trivial bug ttat is going to fixed soon, however in most cases there is no need to dump a full stats. This patch makes possible to select a scope of dumped data by extending current "show stat" to "show stat [<iid> <type> <sid>]": - iid is a proxy id, -1 to dump all proxies - type selects type of dumpable objects: 1 for frontend, 2 for backend, 4 for server, -1 for all types. Values can be ORed, for example: 1+2=3 -> frontend+backend. 1+2+4=7 -> frontend+backend+server. - sid is a service id, -1 to dump everything from the selected proxy. To do this I implemented a new session flag (SN_STAT_BOUND), added three variables in data_ctx.stats (iid, type, sid), modified dumpstats.c and completely revorked the process_uxst_stats: now it waits for a "\n" terminated string, splits args and uses them. BTW: It should be quite easy to add new commands, for example to enable/disable servers, the only problem I can see is a not very lucky config name (*stats* socket). :| During the work I also fixed two bug: - s->flags were not initialized for proto_uxst - missing comma if throttling not enabled (caused by a stupid change in "Implement persistent id for proxies and servers") Other changes: - No more magic type valuse, use STATS_TYPE_FE/STATS_TYPE_BE/STATS_TYPE_SV - Don't memset full s->data_ctx (it was clearing s->data_ctx.stats.{iid/type/sid}, instead initialize stats.sv & stats.sv_st (stats.px and stats.px_st were already initialized) With all that changes it was extremely easy to write a short perl plugin for a perl-enabled net-snmp (also included in this patch). 29385 is my PEN (Private Enterprise Number) and I'm willing to donate the SNMPv2-SMI::enterprises.29385.106.* OIDs for HAProxy if there is nothing assigned already.
2008-03-01 20:42:14 -05:00
/* pid, iid, sid, throttle, lbtot, tracked, type */
[MINOR] switch all stat counters to 64-bit The byte counters have long been 64-bit to avoid overflows. But with several sites nowadays, we see session counters wrap around every 10-days or so. So it was the moment to switch counters to 64-bit, including error and warning counters which can theorically rise as fast as session counters even if in practice there is very low risk. The performance impact should not be noticeable since those counters are only updated once per session. The stats output have been carefully checked for proper types on both 32- and 64-bit platforms.
2009-04-11 14:44:08 -04:00
"%d,%d,0,,%lld,,%d,"
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
/* rate, rate_lim, rate_max, */
"%u,,%u,"
[MEDIUM] Collect & show information about last health check, v3 Collect information about last health check result, including L7 code if possible (for example http or smtp return code) and time took to finish last check. Health check info is provided on both stats pages (html & csv) and logged when a server is marked UP or DOWN. Currently active check are marked with an asterisk, but only in html mode. Currently there are 14 status codes: UNK -> unknown INI -> initializing SOCKERR -> socket error L4OK -> check passed on layer 4, no upper layers testing enabled L4TOUT -> layer 1-4 timeout L4CON -> layer 1-4 connection problem, for example "Connection refused" (tcp rst) or "No route to host" (icmp) L6OK -> check passed on layer 6 L6TOUT -> layer 6 (SSL) timeout L6RSP -> layer 6 invalid response - protocol error L7OK -> check passed on layer 7 L7OKC -> check conditionally passed on layer 7, for example 404 with disable-on-404 L7TOUT -> layer 7 (HTTP/SMTP) timeout L7RSP -> layer 7 invalid response - protocol error L7STS -> layer 7 response error, for example HTTP 5xx
2009-09-23 16:09:24 -04:00
/* check_status, check_code, check_duration */
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
",,,",
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
px->id,
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->nbpend /* or px->totpend ? */, px->be_counters.nbpend_max,
px->beconn, px->be_counters.conn_max, px->fullconn, px->be_counters.cum_conn,
px->be_counters.bytes_in, px->be_counters.bytes_out,
px->be_counters.denied_req, px->be_counters.denied_resp,
px->be_counters.failed_conns, px->be_counters.failed_resp,
px->be_counters.retries, px->be_counters.redispatches,
[MEDIUM] differentiate between generic LB params and map-specific ones Since the introduction of server weights, all load balancing algorithms relied on a pre-computed map. Incidently, quite a bunch of map-specific parameters were used at random places in order to get the number of servers or their total weight. It was not architecturally acceptable that optimizations for the map computation had impact on external parts. For instance, during this cleanup it was found that a backend weight was seen as 1 when only the first backup server is used, whatever its weight. This cleanup consists in differentiating between LB-generic parameters, such as total weights, number of servers, etc... and map-specific ones. The struct proxy has been enhanced in order to make it easier to later support other algorithms. The recount_servers() function now also updates generic values such as total weights so that it's not needed anymore to call recalc_server_map() when weights are needed. This permitted to simplify some code which does not need to know about map internals anymore.
2007-11-15 17:26:18 -05:00
(px->lbprm.tot_weight > 0 || !px->srv) ? "UP" : "DOWN",
[MEDIUM] slowstart: ensure we don't start with a null weight Because of a divide, it was possible to have a null weight during a slowstart, which is pretty annoying, especially with a single server and a long slowstart. Also, fix the way we report the values in the stats page to avoid confusion.
2007-12-02 20:04:00 -05:00
(px->lbprm.tot_weight * px->lbprm.wmult + px->lbprm.wdiv - 1) / px->lbprm.wdiv,
[MEDIUM] differentiate between generic LB params and map-specific ones Since the introduction of server weights, all load balancing algorithms relied on a pre-computed map. Incidently, quite a bunch of map-specific parameters were used at random places in order to get the number of servers or their total weight. It was not architecturally acceptable that optimizations for the map computation had impact on external parts. For instance, during this cleanup it was found that a backend weight was seen as 1 when only the first backup server is used, whatever its weight. This cleanup consists in differentiating between LB-generic parameters, such as total weights, number of servers, etc... and map-specific ones. The struct proxy has been enhanced in order to make it easier to later support other algorithms. The recount_servers() function now also updates generic values such as total weights so that it's not needed anymore to call recalc_server_map() when weights are needed. This permitted to simplify some code which does not need to know about map internals anymore.
2007-11-15 17:26:18 -05:00
px->srv_act, px->srv_bck,
[MINOR] fix several printf formats and missing arguments Last patch revealed a number of mistakes in printf-like calls, mostly int/long mismatches, and a few missing arguments.
2009-04-03 08:49:12 -04:00
px->down_trans, (int)(now.tv_sec - px->last_change),
[MEDIUM] differentiate between generic LB params and map-specific ones Since the introduction of server weights, all load balancing algorithms relied on a pre-computed map. Incidently, quite a bunch of map-specific parameters were used at random places in order to get the number of servers or their total weight. It was not architecturally acceptable that optimizations for the map computation had impact on external parts. For instance, during this cleanup it was found that a backend weight was seen as 1 when only the first backup server is used, whatever its weight. This cleanup consists in differentiating between LB-generic parameters, such as total weights, number of servers, etc... and map-specific ones. The struct proxy has been enhanced in order to make it easier to later support other algorithms. The recount_servers() function now also updates generic values such as total weights so that it's not needed anymore to call recalc_server_map() when weights are needed. This permitted to simplify some code which does not need to know about map internals anymore.
2007-11-15 17:26:18 -05:00
px->srv?be_downtime(px):0,
[STATS] report the number of times each server was selected One user reported that an indicator was missing in the statistics: the number of times each server was selected by load balancing. It is in fact the total number of sessions assigned to a server by the load balancing algorithm. It should directly reflect the weight for "fair" algorithms such as round-robin, since it will not account for persistant connections. It should help a lot tuning each server's weight depending on the load it receives.
2007-12-05 04:34:49 -05:00
relative_pid, px->uuid,
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->be_counters.cum_lbconn, STATS_TYPE_BE,
[MINOR] stats: report max sessions/s and limit in CSV export
2009-05-10 13:01:49 -04:00
read_freq_ctr(&px->be_sess_per_sec),
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->be_counters.sps_max);
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* http response: 1xx, 2xx, 3xx, 4xx, 5xx, other */
if (px->mode == PR_MODE_HTTP) {
int i;
for (i=1; i<6; i++)
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, "%lld,", px->be_counters.p.http.rsp[i]);
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
chunk_printf(&msg, "%lld,", px->be_counters.p.http.rsp[0]);
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
} else {
chunk_printf(&msg, ",,,,,,");
}
[MEDIUM] Decrease server health based on http responses / events, version 3 Implement decreasing health based on observing communication between HAProxy and servers. Changes in this version 2: - documentation - close race between a started check and health analysis event - don't force fastinter if it is not set - better names for options - layer4 support Changes in this version 3: - add stats - port to the current 1.4 tree
2009-12-15 16:31:24 -05:00
/* failed health analyses */
chunk_printf(&msg, ",");
[STATS] report HTTP requests (total and rate) in frontends Now that we support keep-alive, it's important to report a separate counter for requests. Right now it just appears in the CSV output.
2010-02-26 04:05:55 -05:00
/* requests : req_rate, req_rate_max, req_tot, */
chunk_printf(&msg, ",,,");
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
/* errors: cli_aborts, srv_aborts */
chunk_printf(&msg, "%lld,%lld,",
[MEDIUM] stats: split frontend and backend stats It's very annoying that frontend and backend stats are merged because we don't know what we're observing. For instance, if a "listen" instance makes use of a distinct backend, it's impossible to know what the bytes_out means. Some points take care of not updating counters twice if the backend points to the frontend, indicating a "listen" instance. The thing becomes more complex when we try to add support for server side keep-alive, because we have to maintain a pointer to the backend used for last request, and to update its stats. But we can't perform such comparisons anymore because the counters will not match anymore. So in order to get rid of this situation, let's have both frontend AND backend stats in the "struct proxy". We simply update the relevant ones during activity. Some of them are only accounted for in the backend, while others are just for frontend. Maybe we can improve a bit on that later, but the essential part is that those counters now reflect what they really mean.
2011-03-10 17:25:56 -05:00
px->be_counters.cli_aborts, px->be_counters.srv_aborts);
[STATS] count transfer aborts caused by client and by server Often we need to understand why some transfers were aborted or what constitutes server response errors. With those two counters, it is now possible to detect an unexpected transfer abort during a data phase (eg: too short HTTP response), and to know what part of the server response errors may in fact be assigned to aborted transfers.
2010-03-04 14:34:23 -05:00
[MINOR] Collect & provide http response codes received from servers Additional data is provided on both html & csv stats: - html: when passing a mouse over Sessions -> Total (servers, backends) - cvs: by 6 additional fields (hrsp_1xx, hrsp_2xx, hrsp_3xx, hrsp_4xx, hrsp_5xx, hspr_other) Patch inspired by: http://www.formilux.org/archives/haproxy/0910/2528.html http://www.formilux.org/archives/haproxy/0910/2529.html
2009-10-13 15:14:09 -04:00
/* finish with EOL */
chunk_printf(&msg, "\n");
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
}
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 0;
}
[MAJOR] make unix sockets work again with stats The unix protocol handler had not been updated during the last stream_sock changes. This has been done now. There is still a lot of duplicated code between session.c and proto_uxst.c due to the way the session is handled. Session.c relies on the existence of a frontend while it does not exist here. It is easier to see the difference between the stats part (placed in dumpstats.c) and the unix-stream part (in proto_uxst.c). The hijacking function still needs to be dynamically set into the response buffer, and some cleanup is still required, then all those changes should be forward-ported to the HTTP part. Adding support for new keywords should not cause trouble now.
2008-12-07 10:06:43 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px_st = STAT_PX_ST_END;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_END:
if (!(si->applet.ctx.stats.flags & STAT_FMT_CSV)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
chunk_printf(&msg, "</table>");
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (px->cap & PR_CAP_BE && px->srv && (si->applet.ctx.stats.flags & STAT_ADMIN)) {
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
/* close the form used to enable/disable this proxy servers */
chunk_printf(&msg,
"Choose the action to perform on the checked servers : "
"<select name=action>"
"<option value=\"\"></option>"
"<option value=\"disable\">Disable</option>"
"<option value=\"enable\">Enable</option>"
MEDIUM: stats: add support for soft stop/soft start in the admin interface One important missing feature on the web interface is the ability to perform a soft stop/soft start. This is now possible.
2012-06-03 18:22:44 -04:00
"<option value=\"stop\">Soft Stop</option>"
"<option value=\"start\">Soft Start</option>"
MEDIUM: stats: add the ability to kill sessions from the admin interface It was not possible to kill remaining sessions from the admin interface, which is annoying especially when switching to maintenance mode. Now it's possible.
2012-06-03 18:26:23 -04:00
"<option value=\"shutdown\">Kill Sessions</option>"
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"</select>"
MINOR: stats admin: use the backend id instead of its name in the form Proxy ids are unique whereas names can be used several times in the configuration. In order to prevent the ambiguity, the HTML form now provides the backend id instead of its name (the name can still be provided in the POST data).
2012-04-04 06:57:20 -04:00
"<input type=\"hidden\" name=\"b\" value=\"#%d\">"
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
"&nbsp;<input type=\"submit\" value=\"Apply\">"
"</form>",
MINOR: stats admin: use the backend id instead of its name in the form Proxy ids are unique whereas names can be used several times in the configuration. In order to prevent the ambiguity, the HTML form now provides the backend id instead of its name (the name can still be provided in the POST data).
2012-04-04 06:57:20 -04:00
px->uuid);
[MEDIUM] enable/disable servers from the stats web interface Based on a patch provided by Judd Montgomery, it is now possible to enable/disable servers from the stats web interface. This allows to select several servers in a backend and apply the action to them at the same time. Currently, there are 2 known limitations : - The POST data are limited to one packet (don't alter too many servers at a time). - Expect: 100-continue is not supported. (cherry picked from commit 7693948766cb5647ac03b48e782cfee2b1f14491)
2010-10-11 18:14:35 -04:00
}
chunk_printf(&msg, "<p>\n");
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(rep, &msg) == -1)
[MEDIUM] implement the CSV output for the statistics It is now possible to get CSV ouput from the statistics by simply appending ";csv" to the HTTP request sent to get the stats. The fields keep the same ordering as in the HTML page, and a field "pxname" has been prepended at the beginning of the line.
2007-10-17 12:44:57 -04:00
return 0;
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.stats.px_st = STAT_PX_ST_FIN;
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_PX_ST_FIN:
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
return 1;
default:
/* unknown state, we should put an abort() here ! */
return 1;
}
}
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps a complete session state onto the stream intreface's
* read buffer. The data_ctx must have been zeroed first, and the flags
* properly set. The session has to be set in data_ctx.sess.target. It returns
* 0 if the output buffer is full and it needs to be called again, otherwise
* non-zero. It is designed to be called from stats_dump_sess_to_buffer() below.
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_full_sess_to_buffer(struct stream_interface *si)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
{
struct tm tm;
struct chunk msg;
struct session *sess;
extern const char *monthname[12];
char pn[INET6_ADDRSTRLEN];
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
sess = si->applet.ctx.sess.target;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.sess.section > 0 && si->applet.ctx.sess.uid != sess->uniq_id) {
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
/* session changed, no need to go any further */
chunk_printf(&msg, " *** session terminated while we were watching it ***\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.target = NULL;
si->applet.ctx.sess.uid = 0;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 1;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
switch (si->applet.ctx.sess.section) {
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
case 0: /* main status of the session */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.uid = sess->uniq_id;
si->applet.ctx.sess.section = 1;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
/* fall through */
case 1:
chunk_printf(&msg,
"%p: id=%u, proto=%s",
sess,
sess->uniq_id,
sess->listener->proto->name);
MINOR: remove the client/server side distinction in SI addresses Stream interfaces used to distinguish between client and server addresses because they were previously of different types (sockaddr_storage for the client, sockaddr_in for the server). This is not the case anymore, and this distinction is confusing at best and has caused a number of regressions to be introduced in the process of converting everything to full-ipv6. We can now remove this and have a much cleaner code.
2011-09-23 04:54:59 -04:00
switch (addr_to_str(&sess->si[0].addr.from, pn, sizeof(pn))) {
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
case AF_INET:
case AF_INET6:
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, " source=%s:%d\n",
MINOR: remove the client/server side distinction in SI addresses Stream interfaces used to distinguish between client and server addresses because they were previously of different types (sockaddr_storage for the client, sockaddr_in for the server). This is not the case anymore, and this distinction is confusing at best and has caused a number of regressions to be introduced in the process of converting everything to full-ipv6. We can now remove this and have a much cleaner code.
2011-09-23 04:54:59 -04:00
pn, get_host_port(&sess->si[0].addr.from));
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
break;
case AF_UNIX:
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
chunk_printf(&msg, " source=unix:%d\n", sess->listener->luid);
[MINOR] Manage unix socket source field on session dump on sock stats
2010-10-22 10:19:01 -04:00
break;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
default:
/* no more information to print right now */
chunk_printf(&msg, "\n");
break;
}
chunk_printf(&msg,
" flags=0x%x, conn_retries=%d, srv_conn=%p, pend_pos=%p\n",
[MEDIUM] session: move the conn_retries attribute to the stream interface The conn_retries still lies in the session and its initialization depends on the backend when it may not yet be known. Let's first move it to the stream interface.
2010-06-01 03:51:00 -04:00
sess->flags, sess->si[1].conn_retries, sess->srv_conn, sess->pend_pos);
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
chunk_printf(&msg,
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
" frontend=%s (id=%u mode=%s), listener=%s (id=%u)",
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->fe->id, sess->fe->uuid, sess->fe->mode ? "http" : "tcp",
sess->listener ? sess->listener->name ? sess->listener->name : "?" : "?",
sess->listener ? sess->listener->luid : 0);
BUG/MEDIUM: stream_interface: restore get_src/get_dst Commit e164e7a removed get_src/get_dst setting in the stream interfaces but forgot to set it in proto_tcp. Get the feature back because we need it for logging, transparent mode, ACLs etc... We now rely on the stream interface direction to know what syscall to use. One benefit of doing it this way is that we don't use getsockopt() anymore on outgoing stream interfaces nor on UNIX sockets.
2012-05-11 10:16:40 -04:00
si_get_to_addr(&sess->si[0]);
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
switch (addr_to_str(&sess->si[0].addr.to, pn, sizeof(pn))) {
case AF_INET:
case AF_INET6:
chunk_printf(&msg, " addr=%s:%d\n",
pn, get_host_port(&sess->si[0].addr.to));
break;
case AF_UNIX:
chunk_printf(&msg, " addr=unix:%d\n", sess->listener->luid);
break;
default:
/* no more information to print right now */
chunk_printf(&msg, "\n");
break;
}
[MINOR] stats: display "<NONE>" instead of the frontend name when unknown "show sess" should display "<NONE>" instead of the frontend's name as the backend's.
2011-09-02 11:33:05 -04:00
if (sess->be->cap & PR_CAP_BE)
chunk_printf(&msg,
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
" backend=%s (id=%u mode=%s)",
[MINOR] stats: display "<NONE>" instead of the frontend name when unknown "show sess" should display "<NONE>" instead of the frontend's name as the backend's.
2011-09-02 11:33:05 -04:00
sess->be->id,
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
sess->be->uuid, sess->be->mode ? "http" : "tcp");
else
chunk_printf(&msg, " backend=<NONE> (id=-1 mode=-)");
BUG/MEDIUM: stream_interface: restore get_src/get_dst Commit e164e7a removed get_src/get_dst setting in the stream interfaces but forgot to set it in proto_tcp. Get the feature back because we need it for logging, transparent mode, ACLs etc... We now rely on the stream interface direction to know what syscall to use. One benefit of doing it this way is that we don't use getsockopt() anymore on outgoing stream interfaces nor on UNIX sockets.
2012-05-11 10:16:40 -04:00
si_get_from_addr(&sess->si[1]);
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
switch (addr_to_str(&sess->si[1].addr.from, pn, sizeof(pn))) {
case AF_INET:
case AF_INET6:
chunk_printf(&msg, " addr=%s:%d\n",
pn, get_host_port(&sess->si[1].addr.from));
break;
case AF_UNIX:
chunk_printf(&msg, " addr=unix\n");
break;
default:
/* no more information to print right now */
chunk_printf(&msg, "\n");
break;
}
if (sess->be->cap & PR_CAP_BE)
chunk_printf(&msg,
" server=%s (id=%u)",
[MINOR] stats: display "<NONE>" instead of the frontend name when unknown "show sess" should display "<NONE>" instead of the frontend's name as the backend's.
2011-09-02 11:33:05 -04:00
target_srv(&sess->target) ? target_srv(&sess->target)->id : "<none>",
target_srv(&sess->target) ? target_srv(&sess->target)->puid : 0);
else
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
chunk_printf(&msg, " server=<NONE> (id=-1)");
BUG/MEDIUM: stream_interface: restore get_src/get_dst Commit e164e7a removed get_src/get_dst setting in the stream interfaces but forgot to set it in proto_tcp. Get the feature back because we need it for logging, transparent mode, ACLs etc... We now rely on the stream interface direction to know what syscall to use. One benefit of doing it this way is that we don't use getsockopt() anymore on outgoing stream interfaces nor on UNIX sockets.
2012-05-11 10:16:40 -04:00
si_get_to_addr(&sess->si[1]);
MINOR: cli: display the 4 IP addresses and ports on "show sess XXX" I have modified dumpstats.c to show additional information for the show session <id> command on the statistics socket. This will dump the public, frontend, backend, and server ip/tcp addresses and port. We found it useful to have this information available in real time and could not find another way of getting this information.
2012-03-23 13:53:36 -04:00
switch (addr_to_str(&sess->si[1].addr.to, pn, sizeof(pn))) {
case AF_INET:
case AF_INET6:
chunk_printf(&msg, " addr=%s:%d\n",
pn, get_host_port(&sess->si[1].addr.to));
break;
case AF_UNIX:
chunk_printf(&msg, " addr=unix\n");
break;
default:
/* no more information to print right now */
chunk_printf(&msg, "\n");
break;
}
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
chunk_printf(&msg,
" task=%p (state=0x%02x nice=%d calls=%d exp=%s%s)\n",
sess->task,
sess->task->state,
sess->task->nice, sess->task->calls,
sess->task->expire ?
tick_is_expired(sess->task->expire, now_ms) ? "<PAST>" :
human_time(TICKS_TO_MS(sess->task->expire - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>",
task_in_rq(sess->task) ? ", running" : "");
get_localtime(sess->logs.accept_date.tv_sec, &tm);
chunk_printf(&msg,
" task created [%02d/%s/%04d:%02d:%02d:%02d.%06d] (age=%s)\n",
tm.tm_mday, monthname[tm.tm_mon], tm.tm_year+1900,
tm.tm_hour, tm.tm_min, tm.tm_sec, (int)(sess->logs.accept_date.tv_usec),
human_time(now.tv_sec - sess->logs.accept_date.tv_sec, 1));
chunk_printf(&msg,
" si[0]=%p (state=%d flags=0x%02x fd=%d exp=%s, et=0x%03x)\n",
&sess->si[0],
sess->si[0].state,
sess->si[0].flags,
REORG/MINOR: stream_interface: move si->fd to struct connection The socket fd is used only when in socket mode and with a connection.
2012-05-21 10:47:54 -04:00
si_fd(&sess->si[0]),
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->si[0].exp ?
tick_is_expired(sess->si[0].exp, now_ms) ? "<PAST>" :
human_time(TICKS_TO_MS(sess->si[0].exp - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>",
sess->si[0].err_type);
chunk_printf(&msg,
" si[1]=%p (state=%d flags=0x%02x fd=%d exp=%s, et=0x%03x)\n",
&sess->si[1],
sess->si[1].state,
sess->si[1].flags,
REORG/MINOR: stream_interface: move si->fd to struct connection The socket fd is used only when in socket mode and with a connection.
2012-05-21 10:47:54 -04:00
si_fd(&sess->si[1]),
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->si[1].exp ?
tick_is_expired(sess->si[1].exp, now_ms) ? "<PAST>" :
human_time(TICKS_TO_MS(sess->si[1].exp - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>",
sess->si[1].err_type);
chunk_printf(&msg,
" txn=%p (flags=0x%x meth=%d status=%d req.st=%d rsp.st=%d)\n",
&sess->txn, sess->txn.flags, sess->txn.meth, sess->txn.status,
sess->txn.req.msg_state, sess->txn.rsp.msg_state);
chunk_printf(&msg,
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
" req=%p (f=0x%06x an=0x%x i=%d o=%d pipe=%d fwd=%d)\n"
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
" an_exp=%s",
sess->req,
sess->req->flags, sess->req->analysers,
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
sess->req->buf.i, sess->req->buf.o,
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->req->pipe ? sess->req->pipe->data : 0,
sess->req->to_forward,
sess->req->analyse_exp ?
human_time(TICKS_TO_MS(sess->req->analyse_exp - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>");
chunk_printf(&msg,
" rex=%s",
sess->req->rex ?
human_time(TICKS_TO_MS(sess->req->rex - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>");
chunk_printf(&msg,
" wex=%s\n"
MAJOR: http: move buffer->lr to http_msg->next The buffer's pointer <lr> was only used by HTTP parsers which also use a struct http_msg to keep track of the parser's state. We've reached a point where it makes no sense to keep ->lr in the buffer, as the split between buffer and msg is only arbitrary for historical reasons. This change ensures that touching buffers will not impact HTTP messages anymore, making the buffers more content-agnostic. However, it becomes very important not to forget to update msg->next when some data get forwarded or moved (and in general each time buf->p is updated). The new pointer in http_msg becomes relative to buffer->p so that parsing multiple messages becomes easier. It is possible that at one point ->som and ->next will be merged. Note: http_parse_reqline() and http_parse_stsline() have been temporarily modified to know the message starting point in the buffer (->p).
2012-03-05 05:17:50 -05:00
" data=%p p=%d next=%d total=%lld\n",
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->req->wex ?
human_time(TICKS_TO_MS(sess->req->wex - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>",
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
sess->req->buf.data,
(int)(sess->req->buf.p - sess->req->buf.data),
MAJOR: http: move buffer->lr to http_msg->next The buffer's pointer <lr> was only used by HTTP parsers which also use a struct http_msg to keep track of the parser's state. We've reached a point where it makes no sense to keep ->lr in the buffer, as the split between buffer and msg is only arbitrary for historical reasons. This change ensures that touching buffers will not impact HTTP messages anymore, making the buffers more content-agnostic. However, it becomes very important not to forget to update msg->next when some data get forwarded or moved (and in general each time buf->p is updated). The new pointer in http_msg becomes relative to buffer->p so that parsing multiple messages becomes easier. It is possible that at one point ->som and ->next will be merged. Note: http_parse_reqline() and http_parse_stsline() have been temporarily modified to know the message starting point in the buffer (->p).
2012-03-05 05:17:50 -05:00
sess->txn.req.next,
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->req->total);
chunk_printf(&msg,
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
" res=%p (f=0x%06x an=0x%x i=%d o=%d pipe=%d fwd=%d)\n"
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
" an_exp=%s",
sess->rep,
sess->rep->flags, sess->rep->analysers,
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
sess->rep->buf.i, sess->rep->buf.o,
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->rep->pipe ? sess->rep->pipe->data : 0,
sess->rep->to_forward,
sess->rep->analyse_exp ?
human_time(TICKS_TO_MS(sess->rep->analyse_exp - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>");
chunk_printf(&msg,
" rex=%s",
sess->rep->rex ?
human_time(TICKS_TO_MS(sess->rep->rex - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>");
chunk_printf(&msg,
" wex=%s\n"
MAJOR: http: move buffer->lr to http_msg->next The buffer's pointer <lr> was only used by HTTP parsers which also use a struct http_msg to keep track of the parser's state. We've reached a point where it makes no sense to keep ->lr in the buffer, as the split between buffer and msg is only arbitrary for historical reasons. This change ensures that touching buffers will not impact HTTP messages anymore, making the buffers more content-agnostic. However, it becomes very important not to forget to update msg->next when some data get forwarded or moved (and in general each time buf->p is updated). The new pointer in http_msg becomes relative to buffer->p so that parsing multiple messages becomes easier. It is possible that at one point ->som and ->next will be merged. Note: http_parse_reqline() and http_parse_stsline() have been temporarily modified to know the message starting point in the buffer (->p).
2012-03-05 05:17:50 -05:00
" data=%p p=%d next=%d total=%lld\n",
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->rep->wex ?
human_time(TICKS_TO_MS(sess->rep->wex - now_ms),
TICKS_TO_MS(1000)) : "<NEVER>",
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
sess->rep->buf.data,
(int)(sess->rep->buf.p - sess->rep->buf.data),
MAJOR: http: move buffer->lr to http_msg->next The buffer's pointer <lr> was only used by HTTP parsers which also use a struct http_msg to keep track of the parser's state. We've reached a point where it makes no sense to keep ->lr in the buffer, as the split between buffer and msg is only arbitrary for historical reasons. This change ensures that touching buffers will not impact HTTP messages anymore, making the buffers more content-agnostic. However, it becomes very important not to forget to update msg->next when some data get forwarded or moved (and in general each time buf->p is updated). The new pointer in http_msg becomes relative to buffer->p so that parsing multiple messages becomes easier. It is possible that at one point ->som and ->next will be merged. Note: http_parse_reqline() and http_parse_stsline() have been temporarily modified to know the message starting point in the buffer (->p).
2012-03-05 05:17:50 -05:00
sess->txn.rsp.next,
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
sess->rep->total);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 0;
/* use other states to dump the contents */
}
/* end of dump */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.uid = 0;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 1;
}
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps all sessions' states onto the stream intreface's
* read buffer. The data_ctx must have been zeroed first, and the flags
* properly set. It returns 0 if the output buffer is full and it needs
* to be called again, otherwise non-zero. It is designed to be called
* from stats_dump_sess_to_buffer() below.
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_sess_to_buffer(struct stream_interface *si)
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
{
struct chunk msg;
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (unlikely(si->ib->flags & (BF_WRITE_ERROR|BF_SHUTW))) {
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
/* If we're forced to shut down, we might have to remove our
* reference to the last session being dumped.
*/
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
if (si->conn.data_st == STAT_ST_LIST) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!LIST_ISEMPTY(&si->applet.ctx.sess.bref.users)) {
LIST_DEL(&si->applet.ctx.sess.bref.users);
LIST_INIT(&si->applet.ctx.sess.bref.users);
[BUG] fix random memory corruption using "show sess" Commit 8a5c626e73bac905d150185e45110525588d7b4c introduced the sessions dump on the unix socket. This implementation is buggy because it may try to link to the sessions list's head after the last session is removed with a backref. Also, for the LIST_ISEMPTY test to succeed, we have to proceed with LIST_INIT after LIST_DEL.
2009-02-22 09:17:24 -05:00
}
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
}
[MINOR] stats_dump_sess_to_buffer: use buffer_feed_chunk() same as previous patch for this function.
2009-10-03 17:55:14 -04:00
return 1;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
}
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
switch (si->conn.data_st) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INIT:
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
/* the function had not been called yet, let's prepare the
* buffer for a response. We initialize the current session
[BUG] fix random memory corruption using "show sess" Commit 8a5c626e73bac905d150185e45110525588d7b4c introduced the sessions dump on the unix socket. This implementation is buggy because it may try to link to the sessions list's head after the last session is removed with a backref. Also, for the LIST_ISEMPTY test to succeed, we have to proceed with LIST_INIT after LIST_DEL.
2009-02-22 09:17:24 -05:00
* pointer to the first in the global list. When a target
* session is being destroyed, it is responsible for updating
* this pointer. We know we have reached the end when this
* pointer points back to the head of the sessions list.
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
*/
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
LIST_INIT(&si->applet.ctx.sess.bref.users);
si->applet.ctx.sess.bref.ref = sessions.n;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_LIST;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
/* fall through */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_LIST:
[BUG] fix random memory corruption using "show sess" Commit 8a5c626e73bac905d150185e45110525588d7b4c introduced the sessions dump on the unix socket. This implementation is buggy because it may try to link to the sessions list's head after the last session is removed with a backref. Also, for the LIST_ISEMPTY test to succeed, we have to proceed with LIST_INIT after LIST_DEL.
2009-02-22 09:17:24 -05:00
/* first, let's detach the back-ref from a possible previous session */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!LIST_ISEMPTY(&si->applet.ctx.sess.bref.users)) {
LIST_DEL(&si->applet.ctx.sess.bref.users);
LIST_INIT(&si->applet.ctx.sess.bref.users);
[BUG] fix random memory corruption using "show sess" Commit 8a5c626e73bac905d150185e45110525588d7b4c introduced the sessions dump on the unix socket. This implementation is buggy because it may try to link to the sessions list's head after the last session is removed with a backref. Also, for the LIST_ISEMPTY test to succeed, we have to proceed with LIST_INIT after LIST_DEL.
2009-02-22 09:17:24 -05:00
}
/* and start from where we stopped */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
while (si->applet.ctx.sess.bref.ref != &sessions) {
[CLEANUP] Remove unneeded chars allocation Some arrays used to log addresses add some more bytes for ports but this space is never used.
2010-11-01 14:26:02 -04:00
char pn[INET6_ADDRSTRLEN];
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
struct session *curr_sess;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
curr_sess = LIST_ELEM(si->applet.ctx.sess.bref.ref, struct session *, list);
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.sess.target) {
if (si->applet.ctx.sess.target != curr_sess)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
goto next_sess;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
LIST_ADDQ(&curr_sess->back_refs, &si->applet.ctx.sess.bref.users);
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
/* call the proper dump() function and return if we're missing space */
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (!stats_dump_full_sess_to_buffer(si))
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 0;
/* session dump complete */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
LIST_DEL(&si->applet.ctx.sess.bref.users);
LIST_INIT(&si->applet.ctx.sess.bref.users);
si->applet.ctx.sess.target = NULL;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
break;
}
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
"%p: proto=%s",
curr_sess,
curr_sess->listener->proto->name);
MINOR: remove the client/server side distinction in SI addresses Stream interfaces used to distinguish between client and server addresses because they were previously of different types (sockaddr_storage for the client, sockaddr_in for the server). This is not the case anymore, and this distinction is confusing at best and has caused a number of regressions to be introduced in the process of converting everything to full-ipv6. We can now remove this and have a much cleaner code.
2011-09-23 04:54:59 -04:00
switch (addr_to_str(&curr_sess->si[0].addr.from, pn, sizeof(pn))) {
[MINOR] make use of addr_to_str() and get_host_port() to replace many inet_ntop() Many inet_ntop calls were partially right, which was hard to detect given the complex combinations. Some of them were relying on the listener's proto instead of the address itself, which could have been different when dealing with an accept-proxy connection. The new addr_to_str() function does the dirty job and returns the family, which makes it particularly suited to calls from switch/case statements. A large number of if/else statements were removed and the stats output could even be cleaned up in the case of session dump. As a side effect of doing this, the resulting code is smaller by almost 1kB. All changed parts have been tested and provided expected output.
2011-09-04 18:36:48 -04:00
case AF_INET:
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
case AF_INET6:
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
" src=%s:%d fe=%s be=%s srv=%s",
pn,
MINOR: remove the client/server side distinction in SI addresses Stream interfaces used to distinguish between client and server addresses because they were previously of different types (sockaddr_storage for the client, sockaddr_in for the server). This is not the case anymore, and this distinction is confusing at best and has caused a number of regressions to be introduced in the process of converting everything to full-ipv6. We can now remove this and have a much cleaner code.
2011-09-23 04:54:59 -04:00
get_host_port(&curr_sess->si[0].addr.from),
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
curr_sess->fe->id,
[MINOR] stats: display "<NONE>" instead of the frontend name when unknown "show sess" should display "<NONE>" instead of the frontend's name as the backend's.
2011-09-02 11:33:05 -04:00
(curr_sess->be->cap & PR_CAP_BE) ? curr_sess->be->id : "<NONE>",
[MAJOR] session: remove the ->srv pointer from struct session This one has been removed and is now totally superseded by ->target. To get the server, one must use target_srv(&s->target) instead of s->srv now. The function ensures that non-server targets still return NULL.
2011-03-10 10:55:02 -05:00
target_srv(&curr_sess->target) ? target_srv(&curr_sess->target)->id : "<none>"
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
);
break;
case AF_UNIX:
[MINOR] Manage unix socket source field on session dump on sock stats
2010-10-22 10:19:01 -04:00
chunk_printf(&msg,
" src=unix:%d fe=%s be=%s srv=%s",
curr_sess->listener->luid,
curr_sess->fe->id,
[MINOR] stats: display "<NONE>" instead of the frontend name when unknown "show sess" should display "<NONE>" instead of the frontend's name as the backend's.
2011-09-02 11:33:05 -04:00
(curr_sess->be->cap & PR_CAP_BE) ? curr_sess->be->id : "<NONE>",
[MAJOR] session: remove the ->srv pointer from struct session This one has been removed and is now totally superseded by ->target. To get the server, one must use target_srv(&s->target) instead of s->srv now. The function ensures that non-server targets still return NULL.
2011-03-10 10:55:02 -05:00
target_srv(&curr_sess->target) ? target_srv(&curr_sess->target)->id : "<none>"
[MINOR] Manage unix socket source field on session dump on sock stats
2010-10-22 10:19:01 -04:00
);
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
break;
}
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] remove now obsolete ana_state from the session struct This one is not used anymore.
2009-10-04 08:24:59 -04:00
" ts=%02x age=%s calls=%d",
curr_sess->task->state,
[MINOR] show sess: report number of calls to each task For debugging purposes, it can be useful to know how many times each task has been called.
2009-03-28 12:54:35 -04:00
human_time(now.tv_sec - curr_sess->logs.tv_accept.tv_sec, 1),
curr_sess->task->calls);
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
" rq[f=%06xh,i=%d,an=%02xh,rx=%s",
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->req->flags,
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
curr_sess->req->buf.i,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->req->analysers,
curr_sess->req->rex ?
human_time(TICKS_TO_MS(curr_sess->req->rex - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
",wx=%s",
curr_sess->req->wex ?
human_time(TICKS_TO_MS(curr_sess->req->wex - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
",ax=%s]",
curr_sess->req->analyse_exp ?
human_time(TICKS_TO_MS(curr_sess->req->analyse_exp - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
MAJOR: buffer: replace buf->l with buf->{o+i} We don't have buf->l anymore. We have buf->i for pending data and the total length is retrieved by adding buf->o. Some computation already become simpler. Despite extreme care, bugs are not excluded. It's worth noting that msg->err_pos as set by HTTP request/response analysers becomes relative to pending data and not to the beginning of the buffer. This has not been completed yet so differences might occur when outgoing data are left in the buffer.
2012-03-01 12:19:58 -05:00
" rp[f=%06xh,i=%d,an=%02xh,rx=%s",
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->rep->flags,
REORG/MAJOR: extract "struct buffer" from "struct channel" At the moment, the struct is still embedded into the struct channel, but all the functions have been updated to use struct buffer only when possible, otherwise struct channel. Some functions would likely need to be splitted between a buffer-layer primitive and a channel-layer function. Later the buffer should become a pointer in the struct buffer, but doing so requires a few changes to the buffer allocation calls.
2012-07-02 11:01:20 -04:00
curr_sess->rep->buf.i,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->rep->analysers,
curr_sess->rep->rex ?
human_time(TICKS_TO_MS(curr_sess->rep->rex - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
",wx=%s",
curr_sess->rep->wex ?
human_time(TICKS_TO_MS(curr_sess->rep->wex - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
",ax=%s]",
curr_sess->rep->analyse_exp ?
human_time(TICKS_TO_MS(curr_sess->rep->analyse_exp - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
" s0=[%d,%1xh,fd=%d,ex=%s]",
curr_sess->si[0].state,
curr_sess->si[0].flags,
REORG/MINOR: stream_interface: move si->fd to struct connection The socket fd is used only when in socket mode and with a connection.
2012-05-21 10:47:54 -04:00
si_fd(&curr_sess->si[0]),
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->si[0].exp ?
human_time(TICKS_TO_MS(curr_sess->si[0].exp - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
" s1=[%d,%1xh,fd=%d,ex=%s]",
curr_sess->si[1].state,
curr_sess->si[1].flags,
REORG/MINOR: stream_interface: move si->fd to struct connection The socket fd is used only when in socket mode and with a connection.
2012-05-21 10:47:54 -04:00
si_fd(&curr_sess->si[1]),
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
curr_sess->si[1].exp ?
human_time(TICKS_TO_MS(curr_sess->si[1].exp - now_ms),
TICKS_TO_MS(1000)) : "");
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
" exp=%s",
curr_sess->task->expire ?
human_time(TICKS_TO_MS(curr_sess->task->expire - now_ms),
TICKS_TO_MS(1000)) : "");
[OPTIM] task: don't unlink a task from a wait queue when waking it up In many situations, we wake a task on an I/O event, then queue it exactly where it was. This is a real waste because we delete/insert tasks into the wait queue for nothing. The only reason for this is that there was only one tree node in the task struct. By adding another tree node, we can have one tree for the timers (wait queue) and one tree for the priority (run queue). That way, we can have a task both in the run queue and wait queue at the same time. The wait queue now really holds timers, which is what it was designed for. The net gain is at least 1 delete/insert cycle per session, and up to 2-3 depending on the workload, since we save one cycle each time the expiration date is not changed during a wake up.
2009-03-07 11:25:21 -05:00
if (task_in_rq(curr_sess->task))
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, " run(nice=%d)", curr_sess->task->nice);
[MINOR] show sess: report a lot more information about sessions For complex troubleshooting, it's sometimes useful to be able to completely dump all the states and flags related to a session. Now "show sess" will report the stream interfaces and buffers status for each session.
2009-03-28 19:18:14 -04:00
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg, "\n");
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1) {
[BUG] fix random memory corruption using "show sess" Commit 8a5c626e73bac905d150185e45110525588d7b4c introduced the sessions dump on the unix socket. This implementation is buggy because it may try to link to the sessions list's head after the last session is removed with a backref. Also, for the LIST_ISEMPTY test to succeed, we have to proceed with LIST_INIT after LIST_DEL.
2009-02-22 09:17:24 -05:00
/* let's try again later from this session. We add ourselves into
* this session's users so that it can remove us upon termination.
*/
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
LIST_ADDQ(&curr_sess->back_refs, &si->applet.ctx.sess.bref.users);
[MINOR] stats_dump_sess_to_buffer: use buffer_feed_chunk() same as previous patch for this function.
2009-10-03 17:55:14 -04:00
return 0;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
}
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
next_sess:
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.bref.ref = curr_sess->list.n;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
}
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.sess.target) {
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
/* specified session not found */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.sess.section > 0)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
chunk_printf(&msg, " *** session terminated while we were watching it ***\n");
else
chunk_printf(&msg, "Session not found.\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.sess.target = NULL;
si->applet.ctx.sess.uid = 0;
[MINOR] stats socket: add show sess <id> to dump details about a session When trying to spot some complex bugs, it's often needed to access information on stuck sessions, which is quite difficult. This new command helps one get detailed information about a session, with flags, timers, states, etc... The buffer data are not dumped yet.
2010-03-05 11:53:32 -05:00
return 1;
}
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
/* fall through */
default:
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MINOR] stats_dump_sess_to_buffer: use buffer_feed_chunk() same as previous patch for this function.
2009-10-03 17:55:14 -04:00
return 1;
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
}
}
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps all tables' states onto the stream intreface's
* read buffer. The data_ctx must have been zeroed first, and the flags
* properly set. It returns 0 if the output buffer is full and it needs
* to be called again, otherwise non-zero.
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
*/
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
static int stats_table_request(struct stream_interface *si, bool show)
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
{
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
struct session *s = si->conn.data_ctx;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
struct chunk msg;
struct ebmb_node *eb;
int dt;
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
bool skip_entry;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
/*
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
* We have 3 possible states in si->conn.data_st :
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
* - STAT_ST_INIT : the first call
* - STAT_ST_INFO : the proxy pointer points to the next table to
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
* dump, the entry pointer is NULL ;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
* - STAT_ST_LIST : the proxy pointer points to the current table
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
* and the entry pointer points to the next entry to be dumped,
* and the refcount on the next entry is held ;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
* - STAT_ST_END : nothing left to dump, the buffer may contain some
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
* data though.
*/
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (unlikely(si->ib->flags & (BF_WRITE_ERROR|BF_SHUTW))) {
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
/* in case of abort, remove any refcount we might have set on an entry */
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
if (si->conn.data_st == STAT_ST_LIST) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.entry->ref_cnt--;
stksess_kill_if_expired(&si->applet.ctx.table.proxy->table, si->applet.ctx.table.entry);
[MEDIUM] session counters: automatically remove expired entries. When a ref_cnt goes down to zero and the entry is expired, remove it.
2010-08-03 14:34:06 -04:00
}
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
return 1;
}
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
while (si->conn.data_st != STAT_ST_FIN) {
switch (si->conn.data_st) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INIT:
si->applet.ctx.table.proxy = si->applet.ctx.table.target;
if (!si->applet.ctx.table.proxy)
si->applet.ctx.table.proxy = proxy;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.entry = NULL;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INFO;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_INFO:
if (!si->applet.ctx.table.proxy ||
(si->applet.ctx.table.target &&
si->applet.ctx.table.proxy != si->applet.ctx.table.target)) {
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_END;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.table.proxy->table.size) {
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
if (show && !stats_dump_table_head_to_buffer(&msg, si, si->applet.ctx.table.proxy,
si->applet.ctx.table.target))
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
return 0;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.table.target &&
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
s->listener->perm.ux.level >= ACCESS_LVL_OPER) {
/* dump entries only if table explicitly requested */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
eb = ebmb_first(&si->applet.ctx.table.proxy->table.keys);
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
if (eb) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.entry = ebmb_entry(eb, struct stksess, key);
si->applet.ctx.table.entry->ref_cnt++;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_LIST;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
}
}
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.proxy = si->applet.ctx.table.proxy->next;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_LIST:
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
skip_entry = false;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.table.data_type >= 0) {
[MEDIUM] stats: add the ability to dump table entries matching criteria It is now possible to dump some select table entries based on criteria which apply to the stored data. This is enabled by appending the following options to the end of the "show table" statement : data.<data_type> {eq|ne|lt|gt|le|ge} <value> For intance : show table http_proxy data.conn_rate gt 5 show table http_proxy data.gpc0 ne 0 The compare applies to the integer value as it would be displayed, and operates on signed long long integers.
2010-07-18 05:46:20 -04:00
/* we're filtering on some data contents */
void *ptr;
long long data;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
dt = si->applet.ctx.table.data_type;
ptr = stktable_data_ptr(&si->applet.ctx.table.proxy->table,
si->applet.ctx.table.entry,
[MEDIUM] stats: add the ability to dump table entries matching criteria It is now possible to dump some select table entries based on criteria which apply to the stored data. This is enabled by appending the following options to the end of the "show table" statement : data.<data_type> {eq|ne|lt|gt|le|ge} <value> For intance : show table http_proxy data.conn_rate gt 5 show table http_proxy data.gpc0 ne 0 The compare applies to the integer value as it would be displayed, and operates on signed long long integers.
2010-07-18 05:46:20 -04:00
dt);
data = 0;
switch (stktable_data_types[dt].std_type) {
case STD_T_SINT:
data = stktable_data_cast(ptr, std_t_sint);
break;
case STD_T_UINT:
data = stktable_data_cast(ptr, std_t_uint);
break;
case STD_T_ULL:
data = stktable_data_cast(ptr, std_t_ull);
break;
case STD_T_FRQP:
data = read_freq_ctr_period(&stktable_data_cast(ptr, std_t_frqp),
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.proxy->table.data_arg[dt].u);
[MEDIUM] stats: add the ability to dump table entries matching criteria It is now possible to dump some select table entries based on criteria which apply to the stored data. This is enabled by appending the following options to the end of the "show table" statement : data.<data_type> {eq|ne|lt|gt|le|ge} <value> For intance : show table http_proxy data.conn_rate gt 5 show table http_proxy data.gpc0 ne 0 The compare applies to the integer value as it would be displayed, and operates on signed long long integers.
2010-07-18 05:46:20 -04:00
break;
}
/* skip the entry if the data does not match the test and the value */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if ((data < si->applet.ctx.table.value &&
(si->applet.ctx.table.data_op == STD_OP_EQ ||
si->applet.ctx.table.data_op == STD_OP_GT ||
si->applet.ctx.table.data_op == STD_OP_GE)) ||
(data == si->applet.ctx.table.value &&
(si->applet.ctx.table.data_op == STD_OP_NE ||
si->applet.ctx.table.data_op == STD_OP_GT ||
si->applet.ctx.table.data_op == STD_OP_LT)) ||
(data > si->applet.ctx.table.value &&
(si->applet.ctx.table.data_op == STD_OP_EQ ||
si->applet.ctx.table.data_op == STD_OP_LT ||
si->applet.ctx.table.data_op == STD_OP_LE)))
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
skip_entry = true;
[MEDIUM] stats: add the ability to dump table entries matching criteria It is now possible to dump some select table entries based on criteria which apply to the stored data. This is enabled by appending the following options to the end of the "show table" statement : data.<data_type> {eq|ne|lt|gt|le|ge} <value> For intance : show table http_proxy data.conn_rate gt 5 show table http_proxy data.gpc0 ne 0 The compare applies to the integer value as it would be displayed, and operates on signed long long integers.
2010-07-18 05:46:20 -04:00
}
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
if (show && !skip_entry &&
!stats_dump_table_entry_to_buffer(&msg, si, si->applet.ctx.table.proxy,
[MINOR] Break out dumping table This will allow the code to be reused for showing a table filtered by key Signed-off-by: Simon Horman <horms@verge.net.au>
2011-06-15 02:18:45 -04:00
si->applet.ctx.table.entry))
return 0;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.entry->ref_cnt--;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
eb = ebmb_next(&si->applet.ctx.table.entry->key);
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
if (eb) {
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
struct stksess *old = si->applet.ctx.table.entry;
si->applet.ctx.table.entry = ebmb_entry(eb, struct stksess, key);
BUG/MINOR: cli: correctly remove the whole table on "clear table" Joe Price reported that "clear table xxx" sent on the CLI would only clear the last entry. This is true, some code was missing to remove an entry from within the loop, and only the final condition was able to remove an entry. The fix is obvious. No backport is needed.
2012-01-09 05:50:03 -05:00
if (show)
stksess_kill_if_expired(&si->applet.ctx.table.proxy->table, old);
else
stksess_kill(&si->applet.ctx.table.proxy->table, old);
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.entry->ref_cnt++;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
}
[MINOR] More flexible clearing of stick table * Allow clearing of all entries of a table * Allow clearing of all entries of a table that match a data filter
2011-06-15 02:18:49 -04:00
if (show)
stksess_kill_if_expired(&si->applet.ctx.table.proxy->table, si->applet.ctx.table.entry);
else if (!skip_entry && !si->applet.ctx.table.entry->ref_cnt)
stksess_kill(&si->applet.ctx.table.proxy->table, si->applet.ctx.table.entry);
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.table.proxy = si->applet.ctx.table.proxy->next;
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_INFO;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
case STAT_ST_END:
REORG/MEDIUM: stream_interface: move applet->state and private to connection The state and the private pointer are not specific to the applets, since SSL will require exactly both of them. Move them to the connection layer now and rename them. We also now ensure that both are NULL on first call.
2012-05-21 11:09:48 -04:00
si->conn.data_st = STAT_ST_FIN;
[MEDIUM] stats: add "show table [<name>]" to dump a stick-table It is now possible to dump a table's contents with keys, expire, use count, and various data using the command above on the stats socket. "show table" only shows main table stats, while "show table <name>" dumps table contents, only if the socket level is admin.
2010-07-12 11:55:33 -04:00
break;
}
}
return 1;
}
[MINOR] stats: make the data dump function reusable for other purposes The dump_error_line() function was limited to dump error buffers while it's perfectly suitable to dump anything else.
2010-03-05 08:58:26 -05:00
/* print a line of text buffer (limited to 70 bytes) to <out>. The format is :
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
* <2 spaces> <offset=5 digits> <space or plus> <space> <70 chars max> <\n>
* which is 60 chars per line. Non-printable chars \t, \n, \r and \e are
* encoded in C format. Other non-printable chars are encoded "\xHH". Original
* lines are respected within the limit of 70 output chars. Lines that are
* continuation of a previous truncated line begin with "+" instead of " "
* after the offset. The new pointer is returned.
*/
[MINOR] stats: make the data dump function reusable for other purposes The dump_error_line() function was limited to dump error buffers while it's perfectly suitable to dump anything else.
2010-03-05 08:58:26 -05:00
static int dump_text_line(struct chunk *out, const char *buf, int bsize, int len,
int *line, int ptr)
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
{
int end;
unsigned char c;
end = out->len + 80;
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
if (end > out->size)
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
return ptr;
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(out, " %05d%c ", ptr, (ptr == *line) ? ' ' : '+');
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
[MINOR] stats: make the data dump function reusable for other purposes The dump_error_line() function was limited to dump error buffers while it's perfectly suitable to dump anything else.
2010-03-05 08:58:26 -05:00
while (ptr < len && ptr < bsize) {
c = buf[ptr];
[MINOR] show errors: encode backslash as well as non-ascii characters These ones were not properly encoded, causing confusion on the output.
2009-03-12 03:18:33 -04:00
if (isprint(c) && isascii(c) && c != '\\') {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
if (out->len > end - 2)
break;
out->str[out->len++] = c;
[MINOR] show errors: encode backslash as well as non-ascii characters These ones were not properly encoded, causing confusion on the output.
2009-03-12 03:18:33 -04:00
} else if (c == '\t' || c == '\n' || c == '\r' || c == '\e' || c == '\\') {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
if (out->len > end - 3)
break;
out->str[out->len++] = '\\';
switch (c) {
case '\t': c = 't'; break;
case '\n': c = 'n'; break;
case '\r': c = 'r'; break;
case '\e': c = 'e'; break;
[MINOR] show errors: encode backslash as well as non-ascii characters These ones were not properly encoded, causing confusion on the output.
2009-03-12 03:18:33 -04:00
case '\\': c = '\\'; break;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
out->str[out->len++] = c;
} else {
if (out->len > end - 5)
break;
out->str[out->len++] = '\\';
out->str[out->len++] = 'x';
out->str[out->len++] = hextab[(c >> 4) & 0xF];
out->str[out->len++] = hextab[c & 0xF];
}
[MINOR] stats: make the data dump function reusable for other purposes The dump_error_line() function was limited to dump error buffers while it's perfectly suitable to dump anything else.
2010-03-05 08:58:26 -05:00
if (buf[ptr++] == '\n') {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* we had a line break, let's return now */
out->str[out->len++] = '\n';
*line = ptr;
return ptr;
}
}
/* we have an incomplete line, we return it as-is */
out->str[out->len++] = '\n';
return ptr;
}
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
/* This function dumps all captured errors onto the stream intreface's
* read buffer. The data_ctx must have been zeroed first, and the flags
* properly set. It returns 0 if the output buffer is full and it needs
* to be called again, otherwise non-zero.
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
*/
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static int stats_dump_errors_to_buffer(struct stream_interface *si)
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
{
extern const char *monthname[12];
struct chunk msg;
[CLEANUP] stats: make all dump functions only rely on the stream interface This will be needed to move the applet-specific data out of the session.
2011-02-13 09:27:22 -05:00
if (unlikely(si->ib->flags & (BF_WRITE_ERROR|BF_SHUTW)))
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 1;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
BUG/MAJOR: trash must always be the size of a buffer Before it was possible to resize the buffers using global.tune.bufsize, the trash has always been the size of a buffer by design. Unfortunately, the recent buffer sizing at runtime forgot to adjust the trash, resulting in it being too short for content rewriting if buffers were enlarged from the default value. The bug was encountered in 1.4 so the fix must be backported there.
2012-05-16 08:16:48 -04:00
chunk_init(&msg, trash, trashlen);
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (!si->applet.ctx.errors.px) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* the function had not been called yet, let's prepare the
* buffer for a response.
*/
[MINOR] stats: add global event ID and count This counter will help quickly spot whether there are new errors or not. It is also assigned to each capture so that a script can keep trace of which capture was taken when.
2010-12-12 08:00:34 -05:00
struct tm tm;
get_localtime(date.tv_sec, &tm);
chunk_printf(&msg, "Total events captured on [%02d/%s/%04d:%02d:%02d:%02d.%03d] : %u\n",
tm.tm_mday, monthname[tm.tm_mon], tm.tm_year+1900,
tm.tm_hour, tm.tm_min, tm.tm_sec, (int)(date.tv_usec/1000),
error_snapshot_id);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1) {
[MINOR] stats: add global event ID and count This counter will help quickly spot whether there are new errors or not. It is also assigned to each capture so that a script can keep trace of which capture was taken when.
2010-12-12 08:00:34 -05:00
/* Socket buffer full. Let's try again later from the same point */
return 0;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.px = proxy;
si->applet.ctx.errors.buf = 0;
si->applet.ctx.errors.bol = 0;
si->applet.ctx.errors.ptr = -1;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
/* we have two inner loops here, one for the proxy, the other one for
* the buffer.
*/
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
while (si->applet.ctx.errors.px) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
struct error_snapshot *es;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.errors.buf == 0)
es = &si->applet.ctx.errors.px->invalid_req;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
else
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
es = &si->applet.ctx.errors.px->invalid_rep;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
if (!es->when.tv_sec)
goto next;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.errors.iid >= 0 &&
si->applet.ctx.errors.px->uuid != si->applet.ctx.errors.iid &&
es->oe->uuid != si->applet.ctx.errors.iid)
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
goto next;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.errors.ptr < 0) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* just print headers now */
char pn[INET6_ADDRSTRLEN];
struct tm tm;
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
int port;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
get_localtime(es->when.tv_sec, &tm);
[MINOR] stats: don't send empty lines in "show errors" Empty lines indicate end of dump, so it's important not to do that. Send a single space instead.
2010-03-05 11:42:58 -05:00
chunk_printf(&msg, " \n[%02d/%s/%04d:%02d:%02d:%02d.%03d]",
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
tm.tm_mday, monthname[tm.tm_mon], tm.tm_year+1900,
[MINOR] fix several printf formats and missing arguments Last patch revealed a number of mistakes in printf-like calls, mostly int/long mismatches, and a few missing arguments.
2009-04-03 08:49:12 -04:00
tm.tm_hour, tm.tm_min, tm.tm_sec, (int)(es->when.tv_usec/1000));
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
switch (addr_to_str(&es->src, pn, sizeof(pn))) {
case AF_INET:
case AF_INET6:
port = get_host_port(&es->src);
break;
default:
port = 0;
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
switch (si->applet.ctx.errors.buf) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
case 0:
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
" frontend %s (#%d): invalid request\n"
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
" backend %s (#%d)",
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.px->id, si->applet.ctx.errors.px->uuid,
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
(es->oe->cap & PR_CAP_BE) ? es->oe->id : "<NONE>",
(es->oe->cap & PR_CAP_BE) ? es->oe->uuid : -1);
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
break;
case 1:
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
" backend %s (#%d) : invalid response\n"
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
" frontend %s (#%d)",
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.px->id, si->applet.ctx.errors.px->uuid,
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
es->oe->id, es->oe->uuid);
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
break;
}
MEDIUM: http: improve error capture reports A number of important information were missing from the error captures, so let's improve them. Now we also log source port, session flags, transaction flags, message flags, pending output bytes, expected buffer wrapping position, total bytes transferred, message chunk length, and message body length. As such, the output format has slightly evolved and the source address moved to the third line : [08/May/2012:11:14:36.341] frontend echo (#1): invalid request backend echo (#1), server <NONE> (#-1), event #1 src 127.0.0.1:40616, session #4, session flags 0x00000000 HTTP msg state 26, msg flags 0x00000000, tx flags 0x00000000 HTTP chunk len 0 bytes, HTTP body len 0 bytes buffer flags 0x00909002, out 0 bytes, total 28 bytes pending 28 bytes, wrapping at 8030, error at position 7: 00000 GET / /?t=20000 HTTP/1.1\r\n 00026 \r\n [08/May/2012:11:13:13.426] backend echo (#1) : invalid response frontend echo (#1), server local (#1), event #0 src 127.0.0.1:40615, session #1, session flags 0x0000044e HTTP msg state 32, msg flags 0x0000000e, tx flags 0x08200000 HTTP chunk len 0 bytes, HTTP body len 20 bytes buffer flags 0x00008002, out 81 bytes, total 92 bytes pending 11 bytes, wrapping at 7949, error at position 9: 00000 Foo: bar\r\r\n
2012-05-08 05:03:10 -04:00
chunk_printf(&msg,
", server %s (#%d), event #%u\n"
" src %s:%d, session #%d, session flags 0x%08x\n"
" HTTP msg state %d, msg flags 0x%08x, tx flags 0x%08x\n"
" HTTP chunk len %lld bytes, HTTP body len %lld bytes\n"
" buffer flags 0x%08x, out %d bytes, total %lld bytes\n"
" pending %d bytes, wrapping at %d, error at position %d:\n \n",
es->srv ? es->srv->id : "<NONE>", es->srv ? es->srv->puid : -1,
es->ev_id,
pn, port, es->sid, es->s_flags,
es->state, es->m_flags, es->t_flags,
es->m_clen, es->m_blen,
es->b_flags, es->b_out, es->b_tot,
es->len, es->b_wrap, es->pos);
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* Socket buffer full. Let's try again later from the same point */
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 0;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.ptr = 0;
si->applet.ctx.errors.sid = es->sid;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
if (si->applet.ctx.errors.sid != es->sid) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* the snapshot changed while we were dumping it */
[MAJOR] struct chunk rework Add size to struct chunk and simplify the code as there is no longer required to pass sizeof in chunk_printf().
2009-09-27 07:23:20 -04:00
chunk_printf(&msg,
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
" WARNING! update detected on this snapshot, dump interrupted. Please re-check!\n");
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1)
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 0;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
goto next;
}
/* OK, ptr >= 0, so we have to dump the current line */
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
while (si->applet.ctx.errors.ptr < es->len && si->applet.ctx.errors.ptr < sizeof(es->buf)) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
int newptr;
int newline;
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
newline = si->applet.ctx.errors.bol;
newptr = dump_text_line(&msg, es->buf, sizeof(es->buf), es->len, &newline, si->applet.ctx.errors.ptr);
if (newptr == si->applet.ctx.errors.ptr)
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 0;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
MEDIUM: buffers: rename a number of buffer management functions The following renaming took place : 1) buffer input functions buffer_put_block => bi_putblk buffer_put_char => bi_putchr buffer_put_string => bi_putstr buffer_put_chunk => bi_putchk buffer_feed => bi_putstr buffer_feed_chunk => bi_putchk buffer_cut_tail => bi_erase buffer_ignore => bi_fast_delete 2) buffer output functions buffer_get_char => bo_getchr buffer_get_line => bo_getline buffer_get_block => bo_getblk buffer_skip => bo_skip buffer_write => bo_inject 3) buffer input avail/full functions were introduced : bi_avail bi_full
2012-05-07 05:56:55 -04:00
if (bi_putchk(si->ib, &msg) == -1) {
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
/* Socket buffer full. Let's try again later from the same point */
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 0;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.ptr = newptr;
si->applet.ctx.errors.bol = newline;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
};
next:
[REORG] session: move the data_ctx struct to the stream interface's applet This is in fact where those parts belong to. The old data_state was replaced by applet.state and is now initialized when the applet is registered. It's worth noting that the applet does not need to know the session nor the buffer anymore since everything is brought by the stream interface. It is possible that having a separate applet struct would simplify the code but that's not a big deal.
2011-03-10 05:25:07 -05:00
si->applet.ctx.errors.bol = 0;
si->applet.ctx.errors.ptr = -1;
si->applet.ctx.errors.buf++;
if (si->applet.ctx.errors.buf > 1) {
si->applet.ctx.errors.buf = 0;
si->applet.ctx.errors.px = si->applet.ctx.errors.px->next;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
}
/* dump complete */
[MINOR] stats_dump_errors_to_buffer: use buffer_feed_chunk() We can simplify the code in the stats functions using buffer_feed_chunk() instead of buffer_write_chunk(). Let's start with this function. This patch also fixed an issue where we could dump past the end of the capture buffer if it is shorter than the captured request.
2009-10-03 17:49:35 -04:00
return 1;
[MEDIUM] implement error dump on unix socket with "show errors" The new "show errors" command sent on a unix socket will dump all captured request and response errors for all proxies. It is also possible to bound the log to frontends and backends whose ID is passed as an optional parameter. The output provides information about frontend, backend, server, session ID, source address, error type, and error position along with a complete dump of the request or response which has caused the error. If a new error scratches the one currently being reported, then the dump is aborted with a warning message, and processing goes on to next error.
2009-03-04 09:53:18 -05:00
}
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
struct si_applet http_stats_applet = {
.name = "<STATS>", /* used for logging */
.fct = http_stats_io_handler,
MINOR: Add release callback to si_applet
2012-04-02 21:57:53 -04:00
.release = NULL,
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
};
[CLEANUP] dumpstats: make symbols static where possible
2011-06-15 02:18:44 -04:00
static struct si_applet cli_applet = {
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
.name = "<CLI>", /* used for logging */
.fct = cli_io_handler,
MINOR: Add release callback to si_applet
2012-04-02 21:57:53 -04:00
.release = NULL,
[MINOR] stream_interface: make use of an applet descriptor for IO handlers I/O handlers are still delicate to manipulate. They have no type, they're just raw functions which have no knowledge of themselves. Let's have them declared as applets once for all. That way we can have multiple applets share the same handler functions and we can store their names there. When we later need to add more parameters (eg: usage stats), we'll be able to do so in the applets themselves. The CLI functions has been prefixed with "cli" instead of "stats" as it's clearly what is going on there. The applet descriptor in the stream interface should get all the applet specific data (st0, ...) but this will be done in the next patch so that we don't pollute this one too much.
2011-02-13 07:16:36 -05:00
};
[MEDIUM] add support for "show sess" in unix stats socket It is now possible to list all known sessions by issuing "show sess" on the unix stats socket. The format is not much evolved but it is very useful for debugging. The doc has been updated to reflect the new keyword.
2008-12-07 16:29:48 -05:00
[MEDIUM] modularize the global "stats" keyword configuration parser The "stats" keyword already relied on an external parser, let's make use of the new keyword registration mechanism.
2008-07-09 14:12:41 -04:00
static struct cfg_kw_list cfg_kws = {{ },{
{ CFG_GLOBAL, "stats", stats_parse_global },
{ 0, NULL, NULL },
}};
__attribute__((constructor))
static void __dumpstats_module_init(void)
{
cfg_register_keywords(&cfg_kws);
}
[MEDIUM] moved stats and buffer generic functions to new files Neither the primitives used to write data to a buffer, nor the stats dump functions are HTTP-specific anymore. Move them to dedicated files
2007-10-17 11:06:05 -04:00
/*
* Local variables:
* c-indent-level: 8
* c-basic-offset: 8
* End:
*/
Reference in a new issue Copy permalink