Summary:ASTERISK-18267: Channel locks after 'core restart now' and 'core reload'
Reporter:Thomas Arimont (tomaso)Labels:
Date Opened:2011-08-12 08:40:04Date Closed:2011-08-22 15:29:52
Versions: Frequency of
Environment:scientific linux 6Attachments:( 0) channel_locks_reload_during_sip_calls_1psec.log
( 1) channel_locks_restart_during__sip_register_3psec.log
Description:Channel locks after applying commands 'core restart now' and 'core reload' during (normal) SIP-traffic
==> see attached files with results of "core show locks" and gdb thread infos

Using DAHDI timing module.

Comments:By: Richard Mudgett (rmudgett) 2011-08-12 09:26:31.936-0500

The fix for ASTERISK-17760 should also take care of this deadlock.

[^channel_locks_restart_during_ sip_register_3psec.log] is somewhat troubling in that the section for core show locks does not show a thread holding the conlock lock.  Did you edit out other threads there?

By: Thomas Arimont (tomaso) 2011-08-12 09:48:10.353-0500

No, I didn't edit out anything. Do you want me to do the 'restart' case once again? But as far as I understood both cases have the same reason.

By: Thomas Arimont (tomaso) 2011-08-12 10:12:50.150-0500

If I repeat the 'restart' case, the output of 'core show locks' looks exactly the same. There are only these two threads waiting for the same 'conlock'. I guess the third missing thread which holds 'conlock' has been cancelled in the restart procedure ...

By: Thomas Arimont (tomaso) 2011-08-15 11:04:17.721-0500

Thanks, the fix from ASTERISK-17760 seems to solve this problem.