On 30/11/2016 23:46, Alex Balashov wrote:
On Wed, Nov 30, 2016 at 05:42:56PM -0500, Peter
DiSanto wrote:
OK thank you Alex. I will look into your
suggestions next time the problem
arises. Each call does generate one SQL LCR lookup, one CDR INSERT, and
also a timer route that runs 3 queries every few minutes. Once Kamailio
stops responding it will continue that way indefinitely until a process
restart. Will report back. Thanks.
The timer routes run in separate processes,
FYI.
But yeah, it sounds like your SQL DB interactions are the likely source of the problem.
It doesn't have to be the case that these queries get "stuck". If they
become even a little slow, and you don't have sufficient child processes to wait on
them, at 200 CPS that can have consequences. Perhaps profile the queries and/or log slow
queries?
For troubleshooting, when kamailio gets stuck, do a ps and grab all the
pids of kamailio, then take the backtrace from each pid with:
gdb kamailio $PID -batch --eval-command="bt full"
Then we can see what each kamailio project is doing at that moment.
Cheers,
Daniel
--
Daniel-Constantin Mierla
www.twitter.com/miconda --
www.linkedin.com/in/miconda
Kamailio World Conference - May 8-10, 2017 -
www.kamailioworld.com