[asterisk-users] Asterisk 1.4.26.2 died after 80 days uptime

[Olle E. Johansson]

8 feb 2010 kl. 08.37 skrev Steve Totaro:

> On Mon, Feb 8, 2010 at 2:20 AM, Olle E. Johansson <oej at edvina.net> wrote:
>> 
>> 7 feb 2010 kl. 15.09 skrev Per Jessen:
>> 
>>> Thomas Winter wrote:
>>> 
>>>> Hi,
>>>> 
>>>> my Asterisk on debian lenny died after 80 days.
>>>> 
>>>> server kernel: [7572666.186852] asterisk[3673]:
>>>> segfault at 10 ip 7f3b8e90b4aa sp 40bf5f00 error 4 in l
>>>> ibpthread-2.7.so[7f3b8e903000+16000]
>>>> 
>>>> Anything what can be done to find out the reason?
>>> 
>>> My asterisk 1.4.23 also dies about once a month.  I've never been able
>>> to work out why.
>>> 
>> I haven't seen this, but it is definitely something we should try to catch. It could be a memory leak or another type of leak. Any advice from other developers on how to try to catch this?
>> 
>> One thing that would be good would be to get a core dump. There's a document in the /doc directory on how to recompile Asterisk with symbols and force a core dump to happen when we get a crash.
>> 
>> /O
> 
> Just start it with safe_asterisk.
> 
> http://linux.die.net/man/8/safe_asterisk
> 
> Unless my info is out of date, it will kill two birds with one stone.
> Asterisk will restart itself, and you will get a core dump.
There was a reason I referred to the documentation ;-)

YOu will have to recompile it with the DONT_OPTIMIZE variable set so that the core dump actually has meaningful symbols.

/O