haproxy/include/types
Hongbo Long e39683c4d4 BUG/MEDIUM: stream: fix client-fin/server-fin handling
A tcp half connection can cause 100% CPU on expiration.

First reproduced with this haproxy configuration :

  global
      tune.bufsize 10485760
  defaults
      timeout server-fin 90s
      timeout client-fin 90s
  backend node2
      mode tcp
      timeout server 900s
      timeout connect 10s
      server def 127.0.0.1:3333
  frontend fe_api
      mode  tcp
      timeout client 900s
      bind :1990
      use_backend node2

Ie timeout server-fin shorter than timeout server, the backend server
sends data, this package is left in the cache of haproxy, the backend
server continue sending fin package, haproxy recv fin package. this
time the session information is as follows:

  time the session information is as follows:
      0x2373470: proto=tcpv4 src=127.0.0.1:39513 fe=fe_api be=node2
      srv=def ts=08 age=1s calls=3 rq[f=848000h,i=0,an=00h,rx=14m58s,wx=,ax=]
      rp[f=8004c020h,i=0,an=00h,rx=,wx=14m58s,ax=] s0=[7,0h,fd=6,ex=]
      s1=[7,18h,fd=7,ex=] exp=14m58s

  rp has set the CF_SHUTR state, next, the client sends the fin package,

  session information is as follows:
      0x2373470: proto=tcpv4 src=127.0.0.1:39513 fe=fe_api be=node2
      srv=def ts=08 age=38s calls=4 rq[f=84a020h,i=0,an=00h,rx=,wx=,ax=]
      rp[f=8004c020h,i=0,an=00h,rx=1m11s,wx=14m21s,ax=] s0=[7,0h,fd=6,ex=]
      s1=[9,10h,fd=7,ex=] exp=1m11s

  After waiting 90s, session information is as follows:
      0x2373470: proto=tcpv4 src=127.0.0.1:39513 fe=fe_api be=node2
      srv=def ts=04 age=4m11s calls=718074391 rq[f=84a020h,i=0,an=00h,rx=,wx=,ax=]
      rp[f=8004c020h,i=0,an=00h,rx=?,wx=10m49s,ax=] s0=[7,0h,fd=6,ex=]
      s1=[9,10h,fd=7,ex=] exp=? run(nice=0)

  cpu information:
      6899 root      20   0  112224  21408   4260 R 100.0  0.7   3:04.96 haproxy

Buffering is set to ensure that there is data in the haproxy buffer, and haproxy
can receive the fin package, set the CF_SHUTR flag, If the CF_SHUTR flag has been
set, The following code does not clear the timeout message, causing cpu 100%:

stream.c:process_stream:

        if (unlikely((res->flags & (CF_SHUTR|CF_READ_TIMEOUT)) == CF_READ_TIMEOUT)) {
            if (si_b->flags & SI_FL_NOHALF)
            si_b->flags |= SI_FL_NOLINGER;
            si_shutr(si_b);
        }

If you have closed the read, set the read timeout does not make sense.
With or without cf_shutr, read timeout is set:

       if (tick_isset(s->be->timeout.serverfin)) {
           res->rto = s->be->timeout.serverfin;
           res->rex = tick_add(now_ms, res->rto);
       }

After discussion on the mailing list, setting half-closed timeouts the
hard way here doesn't make sense. They should be set only at the moment
the shutdown() is performed. It will also solve a special case which was
already reported of some half-closed timeouts not working when the shutw()
is performed directly at the stream-interface layer (no analyser involved).
Since the stream interface layer cannot know the timeout values, we'll have
to store them directly in the stream interface so that they are used upon
shutw(). This patch does this, fixing the problem.

An easier reproducer to validate the fix is to keep the huge buffer and
shorten all timeouts, then call it under tcploop server and client, and
wait 3 seconds to see haproxy run at 100% CPU :

  global
      tune.bufsize 10485760

  listen px
      bind :1990
      timeout client 90s
      timeout server 90s
      timeout connect 1s
      timeout server-fin 3s
      timeout client-fin 3s
      server def 127.0.0.1:3333

  $ tcploop 3333 L W N20 A P100 F P10000 &
  $ tcploop 127.0.0.1:1990 C S10000000 F
2017-03-21 15:04:43 +01:00
..
acl.h REORG/MAJOR: session: rename the "session" entity to "stream" 2015-04-06 11:23:56 +02:00
action.h MINOR: http: custom status reason. 2017-01-06 11:57:44 +01:00
applet.h MINOR: spoe: Remove SPOE details from the appctx structure 2017-03-09 15:32:55 +01:00
arg.h MINOR: http/conf: store the use_backend configuration file and line for logs 2016-11-25 07:15:09 +01:00
auth.h MAJOR: auth: Change the internal authentication system. 2014-03-17 18:06:06 +01:00
backend.h MEDIUM: backend: add the crc32 hash algorithm for load balancing 2015-01-20 19:48:14 +01:00
capture.h MINOR: capture: extend the captures to support non-header keys 2014-06-13 16:32:48 +02:00
channel.h BUG/MEDIUM: filters: Fix channels synchronization in flt_end_analyze 2017-03-15 19:09:06 +01:00
checks.h MAJOR: check: find out which port to use for health check at run time 2016-09-11 08:12:13 +02:00
cli.h CLEANUP: cli: rename STAT_CLI_* to CLI_ST_* 2016-11-24 16:59:28 +01:00
compression.h MAJOR: filters/http: Rewrite the HTTP compression as a filter 2016-02-09 14:53:15 +01:00
connection.h CLEANUP: connection: completely remove CO_FL_WAKE_DATA 2017-03-19 12:18:27 +01:00
counters.h CLEANUP: counters: move from 3 types to 2 types 2016-11-25 15:03:12 +01:00
dns.h MINOR: dns: implement extra 'hold' timers. 2016-11-09 15:30:47 +01:00
fd.h DEBUG: connection: mark the closed FDs with a value that is easier to detect 2016-11-18 15:00:42 +01:00
filters.h MINOR: filters: Add check_timeouts callback to handle timers expiration on streams 2016-11-21 15:29:58 +01:00
freq_ctr.h [MINOR] freq_ctr: add new types and functions for periods different from 1s 2010-08-10 14:01:09 +02:00
global.h CLEANUP: ssl: move most ssl-specific global settings to ssl_sock.c 2016-12-22 23:26:38 +01:00
hdr_idx.h [BUG] files were missing for hdr_idx in previous commit 2006-12-04 02:20:02 +01:00
hlua.h BUG/MEDIUM: lua: In some case, the return of sample-fetches is ignored (2) 2016-12-14 12:52:47 +01:00
lb_chash.h MINOR: backend: add hash-balance-factor option for hash-type consistent 2016-10-25 20:21:32 +02:00
lb_fas.h MEDIUM: backend: add the 'first' balancing algorithm 2012-02-21 22:27:27 +01:00
lb_fwlc.h [MEDIUM] build: switch ebtree users to use new ebtree version 2009-10-26 21:10:04 +01:00
lb_fwrr.h [MEDIUM] build: switch ebtree users to use new ebtree version 2009-10-26 21:10:04 +01:00
lb_map.h [CLEANUP] proxy: move last lb-specific bits to their respective files 2009-10-03 18:41:18 +02:00
listener.h BUG/MEDIUM: ssl: in bind line, ssl-options after 'crt' are ignored. 2017-03-07 10:42:43 +01:00
log.h CLEANUP: logs: remove unused log format field definitions 2016-08-23 15:25:28 +02:00
mailers.h MINOR: mailers: make it possible to configure the connection timeout 2016-02-20 15:33:06 +01:00
map.h CLEANUP: map: it seems that the map were planed to be chained 2016-03-30 15:41:15 +02:00
obj_type.h CLEANUP: applet: rename struct si_applet to applet 2015-04-23 17:56:16 +02:00
pattern.h MINOR: map: Add regex matching replacement 2016-02-10 23:38:34 +01:00
peers.h MINOR: peers: remove the pointer to the stream 2016-10-31 20:07:01 +01:00
pipe.h [MEDIUM] introduce pipe pools 2009-01-25 13:49:53 +01:00
port_range.h [MEDIUM] add support for binding to source port ranges during connect 2009-06-10 12:23:32 +02:00
proto_http.h BUG/MEDIUM: filters: Fix channels synchronization in flt_end_analyze 2017-03-15 19:09:06 +01:00
proto_udp.h MEDIUM: protocol: add minimalist UDP protocol client 2015-06-13 22:07:35 +02:00
protocol.h CLEANUP: fix inconsistency between fd->iocb, proto->accept and accept() 2016-04-14 11:18:22 +02:00
proxy.h MINOR: server: Add dynamic session cookies. 2017-03-15 11:37:30 +01:00
queue.h REORG/MAJOR: session: rename the "session" entity to "stream" 2015-04-06 11:23:56 +02:00
sample.h BUG/MEDIUM: samples: make smp_dup() always duplicate the sample 2016-08-09 14:03:23 +02:00
server.h MINOR: server: Add dynamic session cookies. 2017-03-15 11:37:30 +01:00
session.h MEDIUM: vars: move the session variables to the session, not the stream 2015-06-19 11:59:02 +02:00
signal.h [MEDIUM] signals: add support for registering functions and tasks 2010-08-27 18:00:40 +02:00
spoe.h MINOR: spoe: Add "send-frag-payload" option in spoe-agent section 2017-03-09 15:32:55 +01:00
ssl_sock.h MEDIUM: boringssl: support native multi-cert selection without bundling 2017-03-02 18:31:05 +01:00
stats.h MEDIUM: stats: Add show json schema 2017-03-14 11:14:03 +01:00
stick_table.h MEDIUM: http: implement http-response track-sc* directive 2016-07-26 14:31:14 +02:00
stream.h MEDIUM: lua: remove Lua struct from session, and allocate it with memory pools 2016-12-21 15:24:56 +01:00
stream_interface.h BUG/MEDIUM: stream: fix client-fin/server-fin handling 2017-03-21 15:04:43 +01:00
task.h DIET/MINOR: task: reduce struct task size by 8 bytes 2013-12-09 16:06:22 +01:00
template.h [CLEANUP] included common/version.h everywhere 2006-06-29 18:54:54 +02:00
vars.h MEDIUM: vars: Add a per-process scope for variables 2016-11-09 22:57:00 +01:00