[asterisk-bugs] [JIRA] (ASTERISK-28560) do_monitor Thread hangs on 99% cpu and doesn't respond

[Eike Maier (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-28560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=248237#comment-248237 ] 

Eike Maier commented on ASTERISK-28560:
---------------------------------------

Thanks for the information. I'm a developer myself and am trying to resolve this problem, too. 
I've been working on this since about two months ago. 

For further information from my side:
I found out that the odbc connector is trying to do a SSL handshake first, then proceed doing a normal connection. When the system is under a bit of load, it seems like the SSL handshake is not proceeded correctly and the connection shut down. Due to the lack of response, the asterisk odbc thread is waiting infinitely, which causes the above mentioned issues.
(See the gdb trace)

After setting the option "SSLMode=Disabled" in odbc.ini, this issue has not surfaced until now.
I'll be watching this issue for a bit and will give a reponse, if it's happening again.

I think this option could be helpful not only in my case, but also in other cases, asterisk realtime is used. 

> do_monitor Thread hangs on 99% cpu and doesn't respond
> ------------------------------------------------------
>
>                 Key: ASTERISK-28560
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-28560
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Channels/chan_sip/Registration
>    Affects Versions: 13.29.0
>         Environment: CentOS 7.5, 2 Cores, 16GB RAM, Gigabit ethernet, XCP hosted VM
>            Reporter: Eike Maier
>
> We have a small Asterisk PBX with about 50 to 70 extensions, 5 simult. calls and are using the queue app. 
> We're using some realtime functions like dynamic realtime for queue, queue member, queue log, moh and voicemail.
> Most of the other config files are used via static realtime. 
> When we're having some calls on our PBX and there are registers at the same time, sometimes the do_monitor thread of asterisk hangs on 99% CPU and doesn't respond to any input. The chan_sip module isn't doing anything after that. Most of the time, only a restart is solving this state.
> This state happens at least 2 times a day and is a bit annoying to reproduce.
> We're using a small modification, to ensure the sippeer table is updated, like via dynamic realtime, but this happens even without the modification.
> I have the output of core show locks and core show threads, a gdb trace could be done the next time, this happens.



--
This message was sent by Atlassian JIRA
(v6.2#6252)