[Asterisk-Dev] Re: [Asterisk-Users] Help Solving Asterisk Lockups

[ewr at erols.com]

>Has anyone been able to build a test case that produces this lockup?  If
>we can at least get to that point, I'd be willing to help track down the
>problem.  I need to be able to reproduce the problem, in order to
>resolve it.

I started this thread on -users.  Sorry to break the threading (wasn't 
subscribed to -dev).

I can't tell you how to recreate this, but I have a machine that is freezing 
up at least once per day.  The debug logs show nothing, so I am assuming I 
will have to use strace or do a backtrace or something along those lines.  I 
have not done either of those before, so any direction you can give me would 
be fantastic.  I saw this readme recently: 
http://bugs.digium.com/view.php?id=5036 .. Would that be the best 
information to provide?

More specific information on our system and the freezes:

The server has a T100P connected to a PRI, and about 50 Polycom IP600 phones 
connected via the local network.  Every couple of hours, Asterisk randomly 
stops responding to all calls, both incoming on the PRI and calls from the 
SIP phones.

When Asterisk stops working I can still connect to it with "asterisk -r". 
Sometimes commands such as "sip show channels" and "zap show channels" show 
the last statuses of when asterisk stopped working.  Other times they do 
nothing.  I am running asterisk at debug level 9, but nothing appears in the 
logs when asterisk stops responding.  Calls to the PRI generate a busy 
signal, and
calls from the SIP phones just time out.  No new messages appear on the 
debug console.

I think that "restart now" fixed the problem once, but I may be imagining 
it.  We usually have to actually kill the asterisk processes and start it 
again.

We were running CVS-HEAD from several weeks ago, but updated to CVS-HEAD on 
Wednesday Aug 24 and are still having the same problem.  We are using 
Slackware 10.1, Realtime (talking to Mysql on the same machine), ldapget, 
app_ldap, rx and txfax, and logging CDR's to mysql.

>Side note: I don't think the cdr_addon_mysql.c code would be doing this,
>unless your not using the patched version and the MySQL server
>disappeared, or you are using the patched version and have the timeout
>set to zero or way to high of a value.

Is this the patch you are referring to: 
http://bugs.digium.com/view.php?id=4953 ?
We have not applied that patch, but mysql is running on the same server (and 
doesn't die or get restarted) so it shouldn't be losing it's MySQL 
connection.... but I'm open to all suggestions!

Eric