[Home]

Summary:ASTERISK-05605: Asterisk crash on heavy load using AGI scripts
Reporter:Eldad Ran (eldadran)Labels:
Date Opened:2005-11-15 02:23:05.000-0600Date Closed:2011-06-07 14:00:53
Priority:CriticalRegression?No
Status:Closed/CompleteComponents:Resources/res_agi
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:( 0) bt.txt
Description:The system just crash when it is loaded, it happended for the 1st time,
Looking at the bt log it looks like a memory issue.


****** ADDITIONAL INFORMATION ******

Please see the enclosed bt trace
Comments:By: Olle Johansson (oej) 2005-11-15 02:52:14.000-0600

Can you repeat the crash in any way?

By: Eldad Ran (eldadran) 2005-11-15 02:55:37.000-0600

It worked for a month with no crashes, I should have the same load in few hours using the same AGI scripts, if it happens again, do you need any actions to be taken?

By: Kevin P. Fleming (kpfleming) 2005-11-15 14:03:44.000-0600

This definitely looks like a heap corruption issue; it will be difficult to find out what actually corrupted the heap, since it may have occurred some time before.

Your best hope of finding the problem will be to run under valgrind or using the astmm memory tracker, but either of those will impact your ability to run in production mode at full performance :-(

By: Eldad Ran (eldadran) 2005-11-15 14:34:58.000-0600

I'm restarting my * daily, can this problem be caused from another process(not *)? or is it * only issue?
I've also experiensed a problem in the event coming from the manager, were it looks like event string are written over another event string.
I'm using astmanproxy, and I'm about to drop it for a while to see if it fucks up the events on heavy load.

By: Tilghman Lesher (tilghman) 2005-11-15 19:56:47.000-0600

The backtrace reveals that you ran out of memory, and the process died for that reason.  You'll need to either increase the available physical memory or increase swap.  There's nothing we can do programatically to avoid the issue of you running out of memory.