|Summary:||ASTERISK-11375: chan_zap pri_dchannel mutex Invalid argument, then segfault|
|Reporter:||Michael FIG (michael-fig)||Labels:|
|Date Opened:||2008-02-04 15:09:25.000-0600||Date Closed:||2011-06-07 14:00:39|
|Environment:||Attachments:||( 0) asterisk-1.4.17-segfault.txt|
|Description:||From time to time, probably once a day under our load, chan_zap crashes Asterisk when trying to acquire a mutex in pri_dchannel.|
We are using VoIP internally, with a Sangoma AFT101 T1 connection to the PSTN.
****** ADDITIONAL INFORMATION ******
chan_zap.c line 9212 (pri_dchannel): Error obtaining mutex: Invalid argument
[My manager API client then returns: Asterisk AMI error tcp_closed: "Connection closed"]
-- Executing [914049928652@internal:1] Set("SIP/531-0831b658", "CALLERID(all)=My Company <3065551212>") in new stack
-- Executing [914049928652@internal:2] Dial("SIP/531-0831b658", "Zap/g0/14049928652||T") in new stack^M
-- Requested transfer capability: 0x00 - SPEECH
-- Called g0/14049928652
-- Zap/2-1 is proceeding passing it to SIP/531-0831b658
|Comments:||By: Jason Parker (jparker) 2008-02-04 15:12:15.000-0600|
We need to see a backtrace. See doc/backtrace.txt
By: Michael FIG (michael-fig) 2008-02-13 16:05:41.000-0600
I've uploaded two backtraces and some GDB analysis of the core dumps (you can ask me more questions about either core file, such as more GDB commands to run, and I'll do that for you).
By: Jason Parker (jparker) 2008-03-20 14:09:43
There have been quite a few fixes lately for random memory stomping.
Could you give the latest svn or 1.4.19-rc3 a try?
By: Jason Parker (jparker) 2008-04-01 13:42:19
Any luck with testing on the latest version?
By: Michael FIG (michael-fig) 2008-04-01 14:43:23
Yes, actually. This issue seems to have gone away for me (no more random crashes in the last few weeks). So, I think it can be closed.
By: Joshua C. Colp (jcolp) 2008-04-01 15:12:55
Closed per reporter.