[ previous ] [ next ] [ threads ]
 To :  Paul Chitescu <paulc@v...>
 From :  Alejandro Ruiz <alitoruiz@g...>
 Subject :  Re: [yate] Crashes
 Date :  Thu, 19 Nov 2009 10:50:00 -0500
Paul any finding on this issue?
Upgrading to 2.1 will help for this?

Regards
Alejandro


On Tue, Nov 17, 2009 at 1:01 PM, Alejandro Ruiz  wrote:

> You can find the core file on http://199.34.121.249/core19146.zip
> let me know if you need me to do a backtrace of the core...
>
> thanks
> alejandro
>
>
>
> On Tue, Nov 17, 2009 at 7:56 AM, Paul Chitescu  wrote:
>
>> Hi!
>>
>> Can you please start Yate with -C -Da on the command line? That will cause
>> it
>> to abort and coredump on the first  so the memory status closer to
>> the
>> problem can be investigated.
>>
>> Note that the backtrace on the thread causing the abort may not be
>> relevant.
>> Most likely there is another thread holding that mutex locked.
>>
>> I suspect a race + deadlock condition caused by the code path that refuses
>> new
>> calls when "full".
>>
>> Regards,
>>
>> Paul
>>
>>
>> On Friday 13 November 2009 09:40:37 pm Alejandro Ruiz wrote:
>> > Here is more information, these lines are from the log.
>> > I am actually using centos 5.3
>> >
>> >
>> >  Not answering H.323 call, full or exiting
>> >  Call disconnect failed - deadlock on call endpoint mutex!
>> >  Call disconnect failed - deadlock on call endpoint mutex!
>> >  Refusing new SIP call, full or exiting
>> >  RTP received SEQ 54827 while current is 32164 [0xa859510]
>> > Yate engine is shutting down with code 128
>> >  Refusing new H.323 call, full or exiting
>> >  16:08.332      ThreadID=0x45683940           assert.cxx(108)   PWLib
>> > Assertion fail: Function pthread_mutex_unlock failed, file tlibthrd
>> > .cxx, line 1893
>> > [...]
>>
>
>



Paul any finding on this issue?
Upgrading to 2.1 will help for this?

Regards
Alejandro


On Tue, Nov 17, 2009 at 1:01 PM, Alejandro Ruiz <alitoruiz@g...> wrote:
You can find the core file on http://199.34.121.249/core19146.zip
let me know if you need me to do a backtrace of the core...

thanks
alejandro



On Tue, Nov 17, 2009 at 7:56 AM, Paul Chitescu <paulc@v...> wrote:
Hi!

Can you please start Yate with -C -Da on the command line? That will cause it
to abort and coredump on the first <FAIL> so the memory status closer to the
problem can be investigated.

Note that the backtrace on the thread causing the abort may not be relevant.
Most likely there is another thread holding that mutex locked.

I suspect a race + deadlock condition caused by the code path that refuses new
calls when "full".

Regards,

Paul


On Friday 13 November 2009 09:40:37 pm Alejandro Ruiz wrote:
> Here is more information, these lines are from the log.
> I am actually using centos 5.3
>
>
> <h323/1516:WARN> Not answering H.323 call, full or exiting
> <FAIL> Call disconnect failed - deadlock on call endpoint mutex!
> <FAIL> Call disconnect failed - deadlock on call endpoint mutex!
> <sip:WARN> Refusing new SIP call, full or exiting
> <WARN> RTP received SEQ 54827 while current is 32164 [0xa859510]
> Yate engine is shutting down with code 128
> <WARN> Refusing new H.323 call, full or exiting
>  16:08.332      ThreadID=0x45683940           assert.cxx(108)   PWLib
> Assertion fail: Function pthread_mutex_unlock failed, file tlibthrd
> .cxx, line 1893
> [...]