Summary:	ASTERISK-14755: Having problems with random crashes
Reporter:	Jim Dickenson (jimdickenson)	Labels:
Date Opened:	2009-09-02 09:25:33	Date Closed:	2009-11-17 07:27:29.000-0600
Priority:	Critical	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:	( 0) core_2009-08-31T21-31-30_notes.txt
Description:	Your pop-up does not have version 1.6.0.14 which is where the error has last occurred. I have had several crashes resulting in core dumps. In looking at where the software stops it seems to be there are multiple problems as there are at least three different places the errors have occurred. I installed 1.6.0.14 after it was released and tried to use the valgrind system to catch a crash. The problem is that when I get about 14 or so concurrent calls going memcheck uses 200% of my multi core multi CPU system and the sound starts to break up. This is unacceptable so I have to not use valgrind when I expect the load to be over 10 concurrent calls. ****** ADDITIONAL INFORMATION ****** I have three malloc_debug and valgrind files as well as a mmlog file but not for when the latest crash occurred. We have written an auto dialer AMI interface that interfaces with our market research interviewing package. We use the AMI so setup queues and move agents in and out of the queues to connect them with answered outbound calls. Except the last one, the previous crashes occurred in 1.6.0.13.
Comments:	By: Jim Dickenson (jimdickenson) 2009-09-02 09:28:08 core 2009-08-31T21-31-30 notes.txt is a backtrace as well as a couple notes from the latest crash By: Jim Dickenson (jimdickenson) 2009-09-02 09:34:58 I have a .gz file with the following in it but it is 2.9MB so it can be retrieved here: http://employees.cfmc.com/dickensn/files.gz malloc_debug.txt malloc_debug.txt.20090831100411 malloc_debug.txt.20090901100230 valgrind.txt valgrind.txt.20090831100411 valgrind.txt.20090901100230 mmlog By: Jim Dickenson (jimdickenson) 2009-09-02 09:37:50 This is asterisk 1.6.0.14 running on a 64 bit version of CentOS 5.3 2.6.18-128.7.1.el5 #1 SMP Mon Aug 24 08:21:56 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux By: David Woolley (davidw) 2009-09-02 11:18:53 Please look at issue ASTERISK-14129.  I think you will find that it is either that problem or one of its children. By: Jim Dickenson (jimdickenson) 2009-09-09 18:15:29 I am having a problem seeing the relation but if you think there is one for version 1.6.0.x where can I get a patch or code that maybe has some fix. I am not doing anything with mp3 files. I am using agents so they do go off and on MOH. Would getting the current 1.6.0 SVN version have the fixes referred to 15109? By: Leif Madsen (lmadsen) 2009-09-10 07:38:25 Yes, latest 1.6.0 SVN would have the 15109 fixes. No releases have been made (yet) with those changes. 1.6.0.15 was just a security release, and 15109 was closed after 1.6.0.14 was created. If you could retest, that would be useful to know. Thanks! By: Jim Dickenson (jimdickenson) 2009-10-23 15:24:19 My client has installed 1.6.0.16-rc2, where I think the fix exists, and is still having random crashes. I actually have three clients getting random crashes. I have tried to use valgrind at one site but once there are more than about 14 users the load added by valgrind makes audio break up so they can run that way. This is on a server class system that has no load problem otherwise. By: carel (carel) 2009-10-27 07:21:28 I have similar issues on 1.6.0.15 on a x86_64. Completely random crashed, *but* seems to happen only a few minutes after having done a reload. First time was a few minutes after a reload and completely idle (brand new installation from digium RPMS) Don't know if this will help. By: Leif Madsen (lmadsen) 2009-11-05 09:25:14.000-0600 Yes, I believe 1.6.0.16-rc2 should have that issue resolved. Hmmm. Can you provide a new backtrace with the latest crashes on the most recent version of Asterisk you can?  Please be sure to enable DONT_OPTIMIZE in menuselect. See doc/backtrace.txt in your Asterisk source for more information. Thanks! By: Jim Dickenson (jimdickenson) 2009-11-05 10:33:52.000-0600 My client has gone to 1.4 as they could not afford all the downtime they were having with all these crashes. I think our company is switching to Asterisk Business Edition as the version we will install at client sites. We are trying to launch a dialer extension to our interviewing software and can not continue fighting with Asterisk crashes. By: Leif Madsen (lmadsen) 2009-11-06 09:24:23.000-0600 OK, I will move this to Acknowledged for now, and will close this if no developer is able to either move this forward, or it remains outstanding for 30 days. By: Leif Madsen (lmadsen) 2009-11-17 07:27:28.000-0600 It is believed the latest set of release candidates for 1.6.0.18 should resolve this issue. Closing for now, but if you're still having an issue, please reopen and provide some additional information. Thanks!