[Serusers] RE: [Serdev] Help needed with core.dumps from SER

[Ricardo Martinez]

Hello Dan.
	First of all thanks for your help on this issue, as i mentioned
before this problem is getting very complicated to me.
	The machine with problems is a "production" machine, so i need to
make the changes in "low load" hours.  But even when the update was made the
crashes are presenting every 4 or 5 days so i need to wait a few day to see
is the problem persist.

	A few last things.
	1.- You mentioned the possibility to be running out of memory.  This
is the output from the "free" command
		
             total       used       free     shared    buffers     cached
Mem:       1030888     993944      36944          0     100892     695612
-/+ buffers/cache:     197440     833448
Swap:      2040244       5188    2035056	

	According to this i have 833M of free memory (the other memory is
cached and buffered).

	2.- I'm downloading SER from CVS with the command : 

	cvs -z3 -d:pserver:anonymous at cvs.ser.berlios.de:/cvsroot/ser co
-rrel_0_9_0 sip_router
	
	To download the changes you introduce in mediaproxy.c i need to run
the same command ????

	3.- How do i compile SER without optimization? 

Thanks again!

Best Regards,
Ricardo Martinez.-

> -----Mensaje original-----
> De: Dan Pascu [mailto:dan at ag-projects.com]
> Enviado el: Miércoles, 20 de Julio de 2005 0:49
> Para: Ricardo Martinez
> CC: 'serdev at lists.iptel.org'; 'serusers at lists.iptel.org'
> Asunto: Re: [Serdev] Help needed with core.dumps from SER
> 
> 
> On Tuesday 19 July 2005 22:00, Ricardo Martinez wrote:
> > Hello.
> > 	On this same issue.  Today i had another crash from my SER.  I'm
> > attaching the gdb output from the core file.   Could this 
> be related to a
> > glibc issue as Dmitry Semyonov pointed? Hope that someone 
> can help me.!
> 
> I don't think it has anything to do with glibc. In mediaproxy 
> there is a place 
> where a memory allocation is not checked. I noticed this 
> before asking you 
> all the info, but I wanted to see all the info I asked for 
> because I wasn't 
> sure that was the only issue. I'm not sure about that even 
> now, because of 
> the weird pointer value (it's 0x5 not 0x0), but that may be 
> because ser was 
> compiled with optimizations and in this case some info is 
> unreliable in gdb.
> Also unless you have very little memory it's highly unlikely 
> that you run out 
> of memory.
> 
> I'll commit the fix to cvs, but I'm still not sure that this 
> memory check is 
> the only issue involved. We can see after that if you still 
> have issues and 
> we can trace them further then if needed. After I commit the 
> fix, if the 
> problem persists, you may need to compile ser without 
> optimizations -O0 to 
> get more reliable information from gdb.
> 
> -- 
> 
> Dan
>