Summary:	ASTERISK-22981: Asterisk crashes while look up AMI action in registered action list
Reporter:	Matteo (mpiazzatnetbug)	Labels:
Date Opened:	2013-12-13 04:00:10.000-0600	Date Closed:	2014-01-21 10:41:56.000-0600
Priority:	Major	Regression?
Status:	Closed/Complete	Components:	Core/ManagerInterface
Versions:	11.2.2	Frequency of Occurrence	One Time
Related Issues:
Environment:	Debian GNU/Linux 6.0	Attachments:	( 0) backtrace.txt
Description:	Tonight I had a segfault on a new asterisk installation. As attachment you can find the complete backtrace. The asterisk was running for one month and there are 200 peers registered on. Into the logsystem i have Dec 13 03:01:16 xxxxx kernel: [2008124.375772] asterisk[1537]: segfault at b5d01174 ip b7609652 sp b4b85784 error 4 in libc-2.11.3.so[b7595000+140000]
Comments:	By: Matt Jordan (mjordan) 2013-12-13 07:52:07.560-0600 Well, that's pretty odd. The manager interface crashed while iterating over the known, registered AMI actions. It's almost as if there was an invalid AMI action in the list. Did you perform any module unloads/reloads prior to this occurring? By: Matteo (mpiazzatnetbug) 2013-12-13 08:46:23.090-0600 It was in the middle of the night, there was no operation in the mean time like reload or sort of. I checked the log of the process that call the manager interface. These are the last two actions sended to the ami. [12-12-2013 23:39:22] c2c rx_command: Action: ExtensionState Context: from-eworks Exten: 3288 ActionID: ewc-1386887957094-3942 Datetime: 2013-12-12 23:39:17 [12-12-2013 23:39:22] c2c command sent to asterisk [13-12-2013 03:01:16] c2c rx_command: Action: ExtensionState Context: from-eworks Exten: 3252 ActionID: ewc-1386900070149-7000 Datetime: 2013-12-13 03:01:10 By: Rusty Newton (rnewton) 2014-01-06 20:41:00.516-0600 Matteo, do you have any debug concerning the state of those extensions at the time the actions were executed? Would you have a log with DEBUG and VERBOSE leading up to the crash that might show what those extensions were doing at the time? By: Matteo (mpiazzatnetbug) 2014-01-07 03:09:40.133-0600 Into the logger.conf file I have messages => notice,warning,error so I don't have more information. Looking into the cdr files and considering it was in the middle of the night I suppose the extension was in "idle" status. Do you suggest me to set verbose into the logger? It's a production pbx do I' dont like the idea to set the debug level. Until now it was a single incident. By: Rusty Newton (rnewton) 2014-01-21 10:41:45.935-0600 We don't have any other similar reports and it only happened once, plus we don't have debug or capability to reproduce. Therefore I'll go ahead and close this out since we can't proceed with diagnosis. I've had a developer look at the backtrace and there is not enough to go on with only that. You may consider running with at least additionally "verbose" logs in the future, "debug" can be intense or generate really big logs if you are running thousands of calls a day. It wouldn't hurt to test briefly with them on and if it didn't cause issues then to try running with them. It can really help when you run into an issue like this. You may also consider upgrading to the latest of the 11 branch for both security fixes and the many various bug fixes since your version. As always I would recommend a test period for it before you roll it into production. Thanks!