Summary:ASTERISK-22014: Crash in libnetsnmpagent
Reporter:Gopalakrishnan N (saigop)Labels:
Date Opened:2013-07-04 17:00:31Date Closed:2013-07-25 10:37:20
Versions:11.2.1 Frequency of
One Time
Environment:CentOS 6.3, Kernel - 2.6.32-279.22.1.el6.x86_64, Asterisk 11.2.1Attachments:( 0) backtrace3.txt
Description:When there was around 80 calls, suddenly asterisk got restarted in 7 seconds of interval, while seeing the core dump we found snmmp message, actually we are monitoring asterisk service, number of channels and active calls via SNMP. Will SNMP be the cause?
Comments:By: Gopalakrishnan N (saigop) 2013-07-04 17:01:36.226-0500

backtrace log as suggested by wiki

By: Matt Jordan (mjordan) 2013-07-04 20:09:28.171-0500

So, it's difficult to see from this backtrace what the problem might be.

The crash isn't actually occurring in Asterisk, it's occurring in {{libnetsnmpagent}}. This could be either a bug in that library, or it could be that Asterisk handed the library invalid input.

Unfortunately, we can't tell if it's the latter case, as Asterisk was optimized when it was compiled. If you can, re-compile Asterisk with the {{DONT_OPTIMIZE}} and {{BETTER_BACKTRACES}} build options - that will show us all of the symbols in the stack when a crash occurs.

If you can reproduce the crash, attach a new backtrace using the instructions on the wiki. Hopefully that will provide enough information to narrow down the problem.

By: Gopalakrishnan N (saigop) 2013-07-05 13:09:58.552-0500

Actually its a production server where right now I can't re-compile, any way that I can check whether the option is enabled in menuselect-tree file?

Please advice.


By: Richard Mudgett (rmudgett) 2013-07-05 19:32:58.827-0500

The CLI command "core show settings" will show what it is running with in the Build Options line.

By: Gopalakrishnan N (saigop) 2013-07-05 19:36:31.883-0500

So what i have is,

PBX Core settings
 Version:                     11.2.1
 Build Options:               LOADABLE_MODULES, BUILD_NATIVE

I think its not enabled.

By: Rusty Newton (rnewton) 2013-07-24 18:12:19.859-0500

Correct. The necessary options are not enabled. Will you be able to re-compile and reproduce the crash?

Hit "Send Back" or use "Enter Feedback" when you reply.

By: Gopalakrishnan N (saigop) 2013-07-25 00:35:09.956-0500

As I said earlier, its an 24/7 production environment, I am not able to recompile, that will affect the production, so time being we disabled SNMP and its keep on running.


By: Gopalakrishnan N (saigop) 2013-07-25 00:36:22.338-0500

After disabling SNMP service in Linux machine there was no crash.

By: Rusty Newton (rnewton) 2013-07-25 10:37:20.255-0500

Closing this out since we don't have a way to reproduce this, and you can't provide additional traces with the needed data to investigate further.

If you are able to provide the requested traces in the future or can describe exact steps that we could use to reproduce the issue in a reasonable way, then we'll re-open the issue.

Feel free to ping a bug marshal in #asterisk-bugs with any questions.

By: dant (dant) 2013-10-25 13:02:00.716-0500

I've seen the same occur within 15 seconds an snmpd restart on a busy system with Asterisk 1.8. I'm looking to reproduce on a test system now.

By: dant (dant) 2013-10-25 14:12:33.024-0500

Appears to be a result of net-snmp bug http://sourceforge.net/p/net-snmp/bugs/2073/ patch at http://net-snmp.svn.sourceforge.net/viewvc/net-snmp/branches/V5-5-patches/agent/mibgroup/agentx/client.c?r1=17856&r2=17855&pathrev=17856

This patch is not present in CentOS 6.

Raised on Redhat Bugzilla at https://bugzilla.redhat.com/show_bug.cgi?id=1023570
Raised on CentOS Bug Tracker at http://bugs.centos.org/view.php?id=6712