Summary:ASTERISK-03437: no MOH to callback agents followed by blocking state
Reporter:David Trcka (dtrcka)Labels:
Date Opened:2005-02-05 05:29:35.000-0600Date Closed:2011-06-07 14:00:38
Versions:Frequency of
Environment:Attachments:( 0) agents.conf
( 1) cli-output-w-comments.txt
( 2) queues.conf
Description:We are using callback agents. When there is only first agent listed in queues.conf logged in, all goes ok. Any other agent logged in causes no MOH for calling party and almost-blocked asterisk. Command "stop now" issued on CLI takes about 20 seconds to finish and causes "Asterisk uncleanly ending (0)".
Comments:By: David Trcka (dtrcka) 2005-02-05 05:35:59.000-0600

note: everything is OK when only agent 211 is logged in.

By: David Trcka (dtrcka) 2005-02-05 05:48:07.000-0600

When i change the order of agents listed in agents.conf, agent 212 works OK and agent 211 causes MOH+block problem.

By: David Trcka (dtrcka) 2005-02-05 07:44:45.000-0600

Problem appears between CVS-D2005. and CVS-D2005., when queue priority was introduced (parameter "weight" in queues.conf).

Latest CVS HEAD and weight parameter commented out: no problem.

By: Mark Spencer (markster) 2005-02-05 08:50:25.000-0600

When it is stuck in this state, can you please attach gdb and provide a full backtrace of all threads (e.g. "thread apply all bt") and attach as a .txt file.  Thanks.

By: David Trcka (dtrcka) 2005-02-05 10:33:15.000-0600

Please provide a step-by-step procedure to obtain this backtrace, I'm not very familiar with this way of debugging.

By: nick (nick) 2005-02-06 10:45:09.000-0600

The easiest instructions are here: http://dev.asteriskdocs.org/index.php/Gdb

There's also a link on the bug guidelines you should have read before filing your bug, but I just looked and half of that link is a mailto:, so I won't fault you too much :).


By: David Trcka (dtrcka) 2005-02-07 07:34:42.000-0600

I'll put the backtrace here later, the machine is in production environment now and I must set up new one for testing environment first.

By: Clod Patry (junky) 2005-02-20 19:18:05.000-0600

dtrcka: if ya do make clean; make valgrind; make install when compiling, if any cores dump occurs, you'll have a core file which will allow you to know what happened exactly.

My opinion, it's always a good idea to have a "make valgrind" in production site too.

By: Mark Spencer (markster) 2005-02-26 01:59:16.000-0600

Need to see thread apply all bt on CVS head here...

By: Mark Spencer (markster) 2005-03-11 00:16:53.000-0600

Closed due to inactivity by bug placer.