Project

General

Profile

Actions
Copy link

Bug #1578

closed
EL VJ

Reload rules under low traffic hangs

Bug #1578: Reload rules under low traffic hangs

Added by Eric Leblond over 10 years ago. Updated about 10 years ago.

Status:
Closed
Priority:
Low
Assignee:
Victor Julien
Target version:
3.0.1
Affected Versions:
Effort:
Difficulty:
Label:

Description

In autofp mode and with a multicore box it is possible the reload of rules get stuck between the two following log messages:

[23912] 16/10/2015 -- 15:44:53 - (detect-engine.c:593) <Notice> (DetectEngineReloadThreads) -- rule reload starting
[23912] 16/10/2015 -- 15:44:58 - (detect-engine.c:747) <Notice> (DetectEngineReloadThreads) -- rule reload complete

On a system with 8 cores, the auto mode of af-packet is starting 8 capture threads and this can cause some threads to get stuck in the DetectEngineReloadThreads function looping and waiting for threads to be ready:

@while (!TmThreadsCheckFlag(tv, THV_RUNNING_DONE)) {
     usleep(100);
}@

Or at line detect-engine.c:703

We should check the thread causing the issue and see how we can implement the timoeout as the delay to reload can be really long.

VJ Updated by Victor Julien over 10 years ago Actions
Copy link
 #1

  • Subject changed from Reload rules under low trafic hangs to Reload rules under low traffic hangs
  • Description updated (diff)

VJ Updated by Victor Julien over 10 years ago Actions
Copy link
 #2

I think this is mostly an issue in workers mode where the threads can block in the packet acquisition for long times.

VJ Updated by Victor Julien about 10 years ago Actions
Copy link
 #3

  • Status changed from New to Closed
  • Assignee changed from Eric Leblond to Victor Julien
  • Target version set to 3.0.1

Seems to be the same as #1716

Actions
Copy link

Also available in: PDF Atom