Summary:ASTERISK-13862: Asterisk crash; stems from chan_agent
Reporter:evandro (evandro)Labels:
Date Opened:2009-03-30 08:41:28Date Closed:2009-06-10 13:01:58
Versions:Frequency of
Environment:Attachments:( 0) gdb.txt
Description:Asterisk crash without a clear reason. These occurs several times for day. I had changed asterisk.conf removing the ";" character from line "dumpcore = yes", but it's not generate the core file. The last lines of full log file was always the following:

[Mar 30 09:02:32] WARNING[8597] codec_gsm.c: Invalid GSM data (1)
[Mar 30 09:02:32] WARNING[8597] translate.c: gsmtolin did not update samples 0

I don't know what information more I can post here...

Any help??
Comments:By: evandro (evandro) 2009-03-31 13:10:33

The crash's did have generate core files, but it's not in /tmp dir as says the backtrace.txt file, it's in my log dir. However, the gdb.txt is attached. Hope it help to solve this problem!

By: Mark Michelson (mmichelson) 2009-03-31 17:49:06

The problem seems to stem from chan_agent. Do you think you could run Asterisk under valgrind and see what happens? See doc/valgrind.txt for instructions. Please notice that running with valgrind may cause the crash not to happen, but valgrind will still report the problem.

By: evandro (evandro) 2009-04-02 07:05:44

Sorry, when I saw this post I already downgrade to asterisk-1.4.19, whose is running on the same computer before the upgrade. This problem was happen in a customer, and needed to be solved, then I was forced to make something.

Until now not crash. One change with upgrade is that calls received in one queue started to be transfered to other queue, and it's generate amount lot of logs like this:

[Mar 31 11:29:20] DEBUG[22478] chan_agent.c: Asked for bridged channel on 'Local/0403@atendimento-cc7d,2'/'Agent/6437', returning 'SIP/6437-08793a50'

other logs that seems strange to me, and was near to the end of all logs when occurs the crash is the follow:

[Mar 31 11:29:20] DEBUG[8686] channel.c: Thread -1605801072 Blocking 'SIP/6401-0876e7b0', already blocked by thread -1606292592 in procedure ast_waitfor_nandfds

By: Leif Madsen (lmadsen) 2009-04-13 13:24:44

evandro: can you reproduce this on a test system? Without the information requested from mmichelson I'll have to close this issue until you can provide it. Thanks!

By: Carsten Haberland (user633) 2009-04-24 11:50:11

I have the same problem, but I don't have a solution. I'm a new asterisk-user.

This is the running system:
- Linux OpenSUSE 10
- Asterisk
- Dahdi-Drivers
- internal: VoIP with Softphones
- external: 4x E1 over Sangoma A104DE
- codecs allow: alaw and ulaw
- codecs disallow: all

There are 1-2 crashs on a day and the server calls 20000 to 30000 numbers over the queue-managing. The crashs are unsteady.

Can you help me? Do you need more informations about the system?

By: Joshua C. Colp (jcolp) 2009-05-06 11:54:26

user633: I would suggest you open a new issue with a backtrace and console output before the crash.

By: Joshua C. Colp (jcolp) 2009-05-06 11:54:46

evandro: Have you been able to get the needed information?

By: evandro (evandro) 2009-05-06 12:06:09

I tryed to reproduce the crash in a test system but don't get it. For while I'm not able to try again, but I pretend to do this very soon as I can.

By: Leif Madsen (lmadsen) 2009-06-10 13:01:58

I'm closing this for now as 'Unable to Reproduce', however as soon as you're able to reproduce this please do re-open this issue and attached the requested information. Thanks!