From marius.zbihlei@1and1.ro Fri Apr  8 11:11:05 2011
From: marius zbihlei <marius.zbihlei@1and1.ro>
To: sr-users@lists.kamailio.org
Subject: Re: [SR-Users] Kamailio stops responding after 10 days or so
Date: Fri, 08 Apr 2011 12:08:41 +0300
Message-ID: <4D9ED099.6020008@1and1.ro>
In-Reply-To: <BANLkTi=o=DWLWF9c9KOtdsO=TEZMhdUp4Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="===============1490795170=="

--===============1490795170==
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit

On 04/07/2011 10:02 PM, Morten Isaksen wrote:
> Hi!
>
> Kamailio 3.0.3.
>
> I have a strange problem with one of our Kamailio servers. This one is
> used for routing (with carrierroute) and to send presence information
> (with pua module)
>
> Once every 10 day or so I get this error and then Kamailio stops
> responding to any SIP packets.
>
> Apr  6 08:05:48 sip-core-1 /usr/local/sbin/kamailio[9186]: WARNING:
> <script>: Failure route - M=INVITE RURI=sip:8615XXXX(a)178.XX.XX.XX
> F=sip:861XXXXX(a)188.120.93.114:1025 T=sip:86155XXXX(a)sip1.uni-tel.dk
> IP=178.XX.XX.XX ID=6de881ec07f9c6494ee589cf208da358(a)10.11.87.206
> Apr  6 08:05:48 sip-core-1 /usr/local/sbin/kamailio[9186]: ERROR:
> carrierroute [cr_func.c:95]: cannot find AVP 'carrier'
> Apr  6 08:05:48 sip-core-1 /usr/local/sbin/kamailio[9186]: ERROR:
> carrierroute [cr_func.c:805]: invalid carrier id -1
> Apr  6 08:05:48 sip-core-1 /usr/local/sbin/kamailio[9186]: ERROR:
> <script>: cr_next_domain failed
>
>    
Hello Morten,

Do you have the carrier avp set in the failure route (it is ok if it's 
set  in normal route it should be visible from there)? Also can you 
paste the lines from failure_route that are around the cr_next_domain . 
My suspicion is that you have some calls that reach failure route, but 
because of some particular cfg lines, they are not correctly routed and 
the daemon just sleeps in some lines of code on each worker.

For replicating the bug and retrieve the line of code where the workers 
block(as Daniel suggested), it would be better if you decrease the 
number of worker and keep the same shm memory config value.

Marius

> Shared memory size is 128M and over halv is free just before the
> error. The server is in production and does handle debug>1 well, so I
> do not have much information in the log files. Private memory is the
> default size.
>
> Any ideas what it could be, or how to investegate further?
>
> I think my next steps would be to increase the private memory og to
> increase children=4 to children=8
>
>    



--===============1490795170==--