[asterisk-dev] Question about hup_handler & deadlocks

[Tim Panton]

On 25 Sep 2006, at 04:10, Jay Hoover wrote:

> Ok, I'll go figure out why I'm getting so many SIGHUP. These are  
> coming in at a pretty steady clip on our production machines,  
> several times an hour. There's probably an unrelated bug somewhere  
> causing this.
>
> As for the deadlock, I agree with Nic that the issue in hup_hanlder  
> should probably get addressed. I'm happy to assist with a patch for  
> this but I don't think I understand Asterisk threading well enough  
> to know which threads is the proper place to perform the actual  
> reload or how to make sure that I don't end up causing other  
> deadlocks in this situation. Any suggestions?
>
> Thanks again,
> Jay
>
> On Sunday 24 September 2006 12:58, Jay Hoover wrote:
> > Thanks, that makes sense. One thing that I don't understand is what
> > situations in normal Asterisk operation would cause a SIGHUP to get
> > sent to the daemon. I'm getting a lot of these deadlocks, and I'm
> > suspicious that there is a problem somewhere else causing me to get
> > an abnormal volume of SIGHUPs. I will do more tracing to track that
> > down, but do you know of anything in normal operation that would
> > cause a large volume of SIGHUPs?
>

The original stacktrace looked to me as if it was in the process of  
exec'ing
an agi script. Any chance your script is sighup'ing it's parent  
process (asterisk),
or the whole process group ?



Tim Panton

www.mexuar.com