[ previous ] [ next ] [ threads ]
 To :  Paul Chitescu <paulc@v...>
 From :  Alejandro Ruiz <alitoruiz@g...>
 Subject :  Re: [yate] h323 crashes
 Date :  Fri, 18 Jun 2010 09:35:08 -0500
Hi,
the maxconn is in 0 (no limit) and the server isn't under heavy load... no
more de than 15 calls on a rate of 1 call every 10 seconds...
I am using the g729 based on intel. manly h323 to h323 on the centos... no
transcoding at all.

the conf you can find it: http://199.34.126.129/yateconf/


Paul, the "full or exiting" is on a centos h323 -> fork -> h323 through a
php script... nothing fancy...

the second log is on a server is on a gentoo that recieves on h323 -> iax,
"the assertion failed on pwlib"

let me see if I can crash it again and collect the trace... and post them

thanks
Alejandro




On Fri, Jun 18, 2010 at 9:28 AM, Paul Chitescu  wrote:

> Martin,
>
> The "full or exiting" is not by configuration. It is caused by the "timeout
> on
> call endpoint mutex" which initiates an Engine::restart(0) in the hope of
> cleaning up the abnormal condition.
>
> What would help in debugging is starting Yate with -C -Da so the FAIL will
> immediately abort() and coredump.
>
> Alejandro, what case is the first crash? h323 -> fork -> h323 ?
>
> Paul
>
>
> On Friday 18 June 2010 05:02:53 pm Martin Provencher wrote:
> > We will need a little bit more information. How much call did you make?
> > Which SDP/Codec do you want to use? What are your *.conf file? How much
> call
> > per seconds/simultaneous call do you have?
> >
> > As I can see from your logs, you have :
> >
> >  Not answering H.323 call, full or exiting
> >  Refusing new H.323 call, full or exiting
> >
> > So, you have more calls on your Yate than what you expected. See
> > h323chan.conf => maxconns. To have more information, you can compile in
> > debug and open the core dump with gdb.
> >
> > Martin
>
>
> >
> >
> > On Fri, Jun 18, 2010 at 9:53 AM, Alejandro Ruiz 
> wrote:
> >
> > > Hi all, here are some new logs about h323 crashes.
> > > There are two different examples on two different servers...
> > >
> > > any clues or updates... ?
> > >
> > > thanks in advance
> > >
> > > Alito
> > >
> > >
> > >
> > >
> > > CENTOS - 3.0.0 runnning php script for rouitng using FORK for fallback.
> > >
> > >  Creating new message dispatching thread (1 running)
> > >  Creating new message dispatching thread (2 running)
> > >  Peer ID failed - timeout on call endpoint mutex owned by 'Call
> > > Router'!
> > >  Formats changed to 'g729'
> > >  Call 'h323/2300' did not get RTP forward from
> 'h323/2301'
> > > target 'h323/yyyyyyyyyyyyy@x...'
> > >  Call 'h323/2300' calling on regular 'fork/619/1' target
> > > 'h323/yyyyyyyyyyyyy@x...'
> > >  Call 'h323/2296' going on 'fork/617/1' to 'h323/2297'
> with
> > > audio data
> > >  Call 'h323/2298' going on 'fork/618/1' to 'h323/2299'
> with
> > > audio data
> > >  Not answering H.323 call, full or exiting
> > >  Not answering H.323 call, full or exiting
> > >  Refusing new H.323 call, full or exiting
> > > PHP: bye!
> > > Supervisor: child 11559 died on signal 11
> > >
> > > ------------------------------------------------
> > >
> > >  Formats changed to 'g729'
> > >  Creating new message dispatching thread (1 running)
> > >  Initial timeout in channel h323/1635 wrapper [0x35a6bd0]
> > >  Creating new message dispatching thread (2 running)
> > >  Peer ID failed - timeout on call endpoint mutex owned by 'Call
> > > Router'!
> > >  Not answering H.323 call, full or exiting
> > >  Formats changed to 'g729'
> > >  Call 'h323/1639' did not get RTP forward from
> 'h323/1642'
> > > target 'h323/yyyyyyyyyyyyy@x...'
> > >  Call 'h323/1639' calling on regular 'fork/441/1' target
> > > 'h323/yyyyyyyyyyyyy@x...'
> > >  Call 'h323/1636' going on 'fork/440/1' to 'h323/1638'
> with
> > > audio data
> > >  Not answering H.323 call, full or exiting
> > >  Refusing new H.323 call, full or exiting
> > > PHP: bye!
> > > Supervisor: child 22863 died on signal 11
> > >
> > > ------------------------------------------------
> > >
> > > GENTOO - 3.0.0 regex recieves on h323 and transcode to iax trunking.
> > >
> > >  Called number (alias) is '50938897307'
> > >  Formats changed to 'g729'
> > > 9:00:20.533        H225 Answer:ac00b7a0       assert.cxx(108)   PWLib
> > > Assertion fail: Invalid array index, file
> > > /usr/src/pwlib/include/ptlib/contain.inl, l
> > > ine 446
> > > Supervisor: child 1167 died on signal 11
> > >
> > > ------------------------------------------------
> > >
> > >  Called number (alias) is '999950938893128'
> > >  Formats changed to 'g729'
> > > 34:54:13.084       H225 Answer:b0011ec0       assert.cxx(108)   PWLib
> > > <80>#^?
> > > Supervisor: child 4716 died on signal 11
> > >
> > >
> >
>



Hi,
the maxconn is in 0 (no limit) and the server isn't under heavy load... no more de than 15 calls on a rate of 1 call every 10 seconds...
I am using the g729 based on intel. manly h323 to h323 on the centos... no transcoding at all.

the conf you can find it: http://199.34.126.129/yateconf/


Paul, the "full or exiting" is on a centos h323 -> fork -> h323 through a php script... nothing fancy...

the second log is on a server is on a gentoo that recieves on h323 -> iax, "the assertion failed on pwlib"

let me see if I can crash it again and collect the trace... and post them

thanks
Alejandro




On Fri, Jun 18, 2010 at 9:28 AM, Paul Chitescu <paulc@v...> wrote:
Martin,

The "full or exiting" is not by configuration. It is caused by the "timeout on
call endpoint mutex" which initiates an Engine::restart(0) in the hope of
cleaning up the abnormal condition.

What would help in debugging is starting Yate with -C -Da so the FAIL will
immediately abort() and coredump.

Alejandro, what case is the first crash? h323 -> fork -> h323 ?

Paul


On Friday 18 June 2010 05:02:53 pm Martin Provencher wrote:
> We will need a little bit more information. How much call did you make?
> Which SDP/Codec do you want to use? What are your *.conf file? How much call
> per seconds/simultaneous call do you have?
>
> As I can see from your logs, you have :
>
> <h323/2303:WARN> Not answering H.323 call, full or exiting
> <WARN> Refusing new H.323 call, full or exiting
>
> So, you have more calls on your Yate than what you expected. See
> h323chan.conf => maxconns. To have more information, you can compile in
> debug and open the core dump with gdb.
>
> Martin


>
>
> On Fri, Jun 18, 2010 at 9:53 AM, Alejandro Ruiz <alitoruiz@g...> wrote:
>
> > Hi all, here are some new logs about h323 crashes.
> > There are two different examples on two different servers...
> >
> > any clues or updates... ?
> >
> > thanks in advance
> >
> > Alito
> >
> >
> >
> >
> > CENTOS - 3.0.0 runnning php script for rouitng using FORK for fallback.
> >
> > <MILD> Creating new message dispatching thread (1 running)
> > <MILD> Creating new message dispatching thread (2 running)
> > <FAIL> Peer ID failed - timeout on call endpoint mutex owned by 'Call
> > Router'!
> > <h323/2301:NOTE> Formats changed to 'g729'
> > <callfork:WARN> Call 'h323/2300' did not get RTP forward from 'h323/2301'
> > target 'h323/yyyyyyyyyyyyy@x...'
> > <callfork:CALL> Call 'h323/2300' calling on regular 'fork/619/1' target
> > 'h323/yyyyyyyyyyyyy@x...'
> > <callfork:NOTE> Call 'h323/2296' going on 'fork/617/1' to 'h323/2297' with
> > audio data
> > <callfork:NOTE> Call 'h323/2298' going on 'fork/618/1' to 'h323/2299' with
> > audio data
> > <h323/2302:WARN> Not answering H.323 call, full or exiting
> > <h323/2303:WARN> Not answering H.323 call, full or exiting
> > <WARN> Refusing new H.323 call, full or exiting
> > PHP: bye!
> > Supervisor: child 11559 died on signal 11
> >
> > ------------------------------------------------
> >
> > <h323/1639:NOTE> Formats changed to 'g729'
> > <MILD> Creating new message dispatching thread (1 running)
> > <yrtp:WARN> Initial timeout in channel h323/1635 wrapper [0x35a6bd0]
> > <MILD> Creating new message dispatching thread (2 running)
> > <FAIL> Peer ID failed - timeout on call endpoint mutex owned by 'Call
> > Router'!
> > <h323/1640:WARN> Not answering H.323 call, full or exiting
> > <h323/1642:NOTE> Formats changed to 'g729'
> > <callfork:WARN> Call 'h323/1639' did not get RTP forward from 'h323/1642'
> > target 'h323/yyyyyyyyyyyyy@x...'
> > <callfork:CALL> Call 'h323/1639' calling on regular 'fork/441/1' target
> > 'h323/yyyyyyyyyyyyy@x...'
> > <callfork:NOTE> Call 'h323/1636' going on 'fork/440/1' to 'h323/1638' with
> > audio data
> > <h323/1641:WARN> Not answering H.323 call, full or exiting
> > <WARN> Refusing new H.323 call, full or exiting
> > PHP: bye!
> > Supervisor: child 22863 died on signal 11
> >
> > ------------------------------------------------
> >
> > GENTOO - 3.0.0 regex recieves on h323 and transcode to iax trunking.
> >
> > <h323/541:INFO> Called number (alias) is '50938897307'
> > <h323/541:NOTE> Formats changed to 'g729'
> > 9:00:20.533        H225 Answer:ac00b7a0       assert.cxx(108)   PWLib
> > Assertion fail: Invalid array index, file
> > /usr/src/pwlib/include/ptlib/contain.inl, l
> > ine 446
> > Supervisor: child 1167 died on signal 11
> >
> > ------------------------------------------------
> >
> > <h323/8630:INFO> Called number (alias) is '999950938893128'
> > <h323/8630:NOTE> Formats changed to 'g729'
> > 34:54:13.084       H225 Answer:b0011ec0       assert.cxx(108)   PWLib
> > <80>#^?
> > Supervisor: child 4716 died on signal 11
> >
> >
>