mirror of
https://github.com/NLnetLabs/unbound.git
synced 2025-12-20 23:00:56 -05:00
- Increase default to num-queries-per-thread: 2048, when unbound is
compiled with libevent. It makes saturation of the task queue more resource intensive and less practical. Thanks to Shiming Liu, Network and Information Security Lab, Tsinghua University for the report.
This commit is contained in:
parent
9c99b404a1
commit
fe835f9d52
4 changed files with 11 additions and 4 deletions
|
|
@ -1,3 +1,10 @@
|
||||||
|
16 April 2025: Wouter
|
||||||
|
- Increase default to `num-queries-per-thread: 2048`, when unbound is
|
||||||
|
compiled with libevent. It makes saturation of the task queue more
|
||||||
|
resource intensive and less practical. Thanks to Shiming Liu,
|
||||||
|
Network and Information Security Lab, Tsinghua University for the
|
||||||
|
report.
|
||||||
|
|
||||||
11 April 2025: Yorgos
|
11 April 2025: Yorgos
|
||||||
- Merge #1265: Fix WSAPoll.
|
- Merge #1265: Fix WSAPoll.
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -163,7 +163,7 @@ server:
|
||||||
# msg-cache-slabs: 4
|
# msg-cache-slabs: 4
|
||||||
|
|
||||||
# the number of queries that a thread gets to service.
|
# the number of queries that a thread gets to service.
|
||||||
# num-queries-per-thread: 1024
|
# num-queries-per-thread: 2048
|
||||||
|
|
||||||
# if very busy, 50% queries run to completion, 50% get timeout in msec
|
# if very busy, 50% queries run to completion, 50% get timeout in msec
|
||||||
# jostle-timeout: 200
|
# jostle-timeout: 200
|
||||||
|
|
|
||||||
|
|
@ -268,7 +268,7 @@ The number of queries that every thread will service simultaneously.
|
||||||
If more queries arrive that need servicing, and no queries can be jostled out
|
If more queries arrive that need servicing, and no queries can be jostled out
|
||||||
(see \fIjostle\-timeout\fR), then the queries are dropped. This forces
|
(see \fIjostle\-timeout\fR), then the queries are dropped. This forces
|
||||||
the client to resend after a timeout; allowing the server time to work on
|
the client to resend after a timeout; allowing the server time to work on
|
||||||
the existing queries. Default depends on compile options, 512 or 1024.
|
the existing queries. Default depends on compile options, 512 or 2048.
|
||||||
.TP
|
.TP
|
||||||
.B jostle\-timeout: \fI<msec>
|
.B jostle\-timeout: \fI<msec>
|
||||||
Timeout used when the server is very busy. Set to a value that usually
|
Timeout used when the server is very busy. Set to a value that usually
|
||||||
|
|
@ -280,7 +280,7 @@ service by slow queries or high query rates. Default 200 milliseconds.
|
||||||
The effect is that the qps for long-lasting queries is about
|
The effect is that the qps for long-lasting queries is about
|
||||||
(numqueriesperthread / 2) / (average time for such long queries) qps.
|
(numqueriesperthread / 2) / (average time for such long queries) qps.
|
||||||
The qps for short queries can be about (numqueriesperthread / 2)
|
The qps for short queries can be about (numqueriesperthread / 2)
|
||||||
/ (jostletimeout in whole seconds) qps per thread, about (1024/2)*5 = 2560
|
/ (jostletimeout in whole seconds) qps per thread, about (2048/2)*5 = 5120
|
||||||
qps by default.
|
qps by default.
|
||||||
.TP
|
.TP
|
||||||
.B delay\-close: \fI<msec>
|
.B delay\-close: \fI<msec>
|
||||||
|
|
|
||||||
|
|
@ -155,7 +155,7 @@ config_create(void)
|
||||||
# else
|
# else
|
||||||
/* libevent can use many sockets */
|
/* libevent can use many sockets */
|
||||||
cfg->outgoing_num_ports = 4096;
|
cfg->outgoing_num_ports = 4096;
|
||||||
cfg->num_queries_per_thread = 1024;
|
cfg->num_queries_per_thread = 2048;
|
||||||
# endif
|
# endif
|
||||||
cfg->outgoing_num_tcp = 10;
|
cfg->outgoing_num_tcp = 10;
|
||||||
cfg->incoming_num_tcp = 10;
|
cfg->incoming_num_tcp = 10;
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue