Dear all,
We're experiencing some (what appear to be) deadlocks using
asterisk-1.4.35 (the problem also occurred on 1.4.24). Some of the
symptoms are that new SIP calls cannot be established and when running
"sip show channels" from the CLI, the CLI stops responding to any
further commands. The symptoms are in fact very similar to
https://issues.asterisk.org/view.php?id=15349. A possibly significant
difference is that once asterisk hangs and I try to stop or kill it,
asterisk will remain as a defunct process and I have to reboot the
machine to get things working again.
Two servers are involved, both running asterisk 1.4.35, both have a
TC400B card used for G729 transcoding and both are equipped with a
Sangoma A104dm card connected to 4x ISDN-30. Server A is the main server
(holds all phone registrations, does all the monitoring/recording, all
outgoing calls originate here and all incoming calls terminate here).
Server B acts as a SIP trunk in a neighboring country (connected by a
stable, dedicated and private link).
Every once in a while (about 2-4 times a week), either one of the
servers stops setting up new SIP channels, something we haven't
experienced before on other sites (only significant difference is that
the other sites use an older version of asterisk and lack a transcoder
card).
I've managed to get the output of "core show locks" (available
here:
http://pastebin.com/S870RSi3), unfortunately, I haven't been able to
collect a backtrace yet.
Now down to my question: What I'd like to know is how to properly debug
this issue further.
On a side-note, there are a couple of problems with debugging this
issue: a) We've been unable to reproduce it locally. Actively trying to
reproduce it on the production machine is a no-go due to contracts with
the customers. And b) once the system does hang, we have very little
time to poke around (again, because of contracts), the system has to be
up and running as soon as possible.
Kind regards,
Ingmar Steen
Teleknowledge
