Summary:ASTERISK-01844: MeetMe causes PBX to lock up (g729 possibly involved)
Reporter:pi (pi)Labels:
Date Opened:2004-06-16 17:54:44Date Closed:2011-06-07 14:10:39
Versions:Frequency of
Environment:Attachments:( 0) backtrace.txt
( 1) backtrace2.txt
( 2) errorlog.txt
( 3) thread.txt
Description:Occasionally when exiting a MeetMe conference with Budgetone SIP phones using the g.729 codec, the PBX goes "catatonic" and refuses to respond to any further requests. All current ZAP + SIP bridged calls remain active. "show channels" from an asterisk -r session returns empty. "show channels" in the pbx console locks the CLI up.

I did ctrl-\ to force a coredump and ran gdb against it. Attached is error log output before a ctrl-\. A backtrace will be attached after submission.


I have not tested meetme heavily with another codec. This didn't happen when we were using iLBC codec, but it isn't immediately reproducible with g729 either. Normal ZAP, IAX and SIP calls without MeetMe are OK.
Comments:By: Mark Spencer (markster) 2004-06-16 20:19:15

Curiously, I think this is the same bug as the PRI "ring request" error since both involve the mismatch of codecs.

By: Mark Spencer (markster) 2004-06-16 21:08:52

Can you duplicate this bug?  If so, find me on IRC so that I can login and do a backtrace and get this puppy knocked out once and for all.

By: Mark Spencer (markster) 2004-06-17 09:56:06

Assuming you were using Zap PRI in some capacity (which it appears you were), some updates last night may fix up your issues.

By: pi (pi) 2004-06-17 09:58:36

No, just TDM400 and FXO interfaces.

By: Brian West (bkw918) 2004-06-17 13:55:42

Closed pending reproduction.

By: pi (pi) 2004-06-18 13:03:57

Reproduced issue. Backtrace is attached.

By: Mark Spencer (markster) 2004-06-18 14:51:11

Go ahead and install valgrind, then find me on IRC and we'll see if we can't track this down.  The backtrace is not consistent with a traditional deadlock because nobody actually *has* any locks here that are in contention, so presumably there is somewhere that a lock is being grabbed and not released.  Might even have to do some thread debugging.

By: Mark Spencer (markster) 2004-06-24 00:57:10

What's the status here?  can you cvs update and see if this bug still exists?

By: Mark Spencer (markster) 2004-06-27 21:54:45

User seems to have lost interest and I'm unable to duplicate.

By: pi (pi) 2004-06-27 23:02:57

Sorry - I was away from my computer for a good 4 days (registering for college courses). I had a report that the lockup occured while I was gone. The issue appears to be with the Budgetone 1.0.4.xx firmware that shipped with our phones (or how my co-worker is transfering - i've never personally caused the lockup (even with her phones), only her while she's been using her phones.) I'm sorry I let this bug languish for the time I did, but i'm still interested in debugging this. Asterisk CVS-HEAD-06/18/04-12:14:43 was the version running at the time of the reported lockup. I am now updating to CVS-HEAD-06/27/04-21:50:31 and will report back on next lockup.

edited on: 06-27-04 22:50

By: Mark Spencer (markster) 2004-06-27 23:28:19

Then find me on IRC when it's stuck in that state and I'll debug it.  I won't be able to do anything else without having it stuck *and* being able to attach gdb and get a meaningful feedback.

By: Brian West (bkw918) 2004-07-01 22:55:06

closed pending recreation.