[asterisk-bugs] [JIRA] (ASTERISK-22833) channel: Channel reference leak

[Rusty Newton (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-22833?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=214133#comment-214133 ] 

Rusty Newton commented on ASTERISK-22833:
-----------------------------------------

Matt Jordan analyzed your provided refs file, but it still turned out to be benign, so still nothing else to go on unfortunately.

He did notice that you were using res_timing_pthread.so, which is extended support and is known to have issues. He and others recommend using res_timing_timerfd.so if possible. [You can find additional info on timing interfaces here|https://wiki.asterisk.org/wiki/display/AST/Timing+Interfaces].

I re-tested with your described conditions and dialplan using res_timing_pthread.so to see what would happen. After around 1200 call attempts I still didn't see any growth in file descriptors, stuck channels, or anything strange for that matter.

I suppose that in the scenarios where you reproduce the issue, there is likely something the channel is doing that isn't happening in my test.

I'm going to close this out for now as Cannot Reproduce. If you can provide an exact configuration scenario to reproduce the issue then we can re-open, but otherwise the refs info and other debug hasn't been helpful to any of the developers looking at it so far. At least the issue is logged here, so if someone else reports a similar issue then we have more knowledge on it. I know this is frustrating, but we don't have anything else to go on.
                
> channel: Channel reference leak
> -------------------------------
>
>                 Key: ASTERISK-22833
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-22833
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Core/Channels
>    Affects Versions: 11.5.1, 11.6.0
>         Environment: centos
>            Reporter: Mitch Rodrigues
>            Assignee: Rusty Newton
>            Severity: Critical
>         Attachments: extensions.txt, fd_after_1000.txt, fd_at_idle.txt, refcounter_output.txt, refs, refs.txt, sip.txt
>
>
> fd pipes linger from channel internals after call has been terminated. Seems to to also leave around alot of anon_inodes when doing lsof 
>  

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.asterisk.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira