[OpenSIPS-Users] Opensips Late reply.

[qasimakhan at gmail.com]

@Daniel, I know i dont need 1 child per request but i did it because it
might keep delays short as it has more threads to work with. I have limited
input call origination's from mi_datagram's UDP interface. Also maximum
number of calls are capped using dialog profiling.


Also i am using opensips 1.11 and i have got a few more logs to share. This
time i deliberately put maximum load on opensips and let it crash, Here are
is some info collected from core dumps:

*Log File:*
2017-04-05T17:45:04.986687+05:00 sip01 /usr/local/sbin/opensips[29092]:
Total number of dropped replies = 0
2017-04-05T17:45:04.986764+05:00 sip01 /usr/local/sbin/opensips[29344]:
Total number of dropped replies = 0
2017-04-05T17:45:04.987717+05:00 sip01 /usr/local/sbin/opensips[29344]:
Real SHMEM used size is 126444552
2017-04-05T17:45:04.988274+05:00 sip01 /usr/local/sbin/opensips[29344]:
Free SHMEM available is 947297272
2017-04-05T17:45:04.988425+05:00 sip01 /usr/local/sbin/opensips[29344]: The
total number of SHMEM fragments is 99819

2017-04-05T17:45:35.519928+05:00 sip01 /usr/local/sbin/opensips[28681]:
INFO:core:handle_sigs: child process 28887 exited by a signal 6
2017-04-05T17:45:35.730406+05:00 sip01 /usr/local/sbin/opensips[28681]:
INFO:core:handle_sigs: core was generated
2017-04-05T17:45:35.730879+05:00 sip01 /usr/local/sbin/opensips[28681]:
INFO:core:handle_sigs: terminating due to SIGCHLD

*gdb bt full:*
#5  0x00007f3a81d17f7e in mi_uac_dlg_hdl (t=<optimized out>,
type=<optimized out>, ps=0x7ffe99c7eeb0) at mi.c:378
        mi_hdl = 0x7f3a44b04b18
        rpl_tree = 0x7f3a825f9d60
        text = {s = 0x3 <error: Cannot access memory at address 0x3>, len =
0}
        __FUNCTION__ = "mi_uac_dlg_hdl"


-Qasim


On Wed, Apr 5, 2017 at 5:53 PM, Daniel Zanutti <daniel.zanutti at gmail.com>
wrote:

> Hi Qasim
>
> How did you limit CPS? Because i have a similar scenario (150cps) but i
> set children to 20 or 24, never 200. You don't need 1 children per request.
>
> On Wed, Apr 5, 2017 at 9:44 AM, qasimakhan at gmail.com <qasimakhan at gmail.com
> > wrote:
>
>> Hi,
>>
>> I have this scenario where i originate calls from mi_datagram and the
>> calls are cancelled as soon as it starts ringing. The problem i am facing
>> is that are running for a few minutes the response from opensips becomes
>> slow i.e. it send packets back to far end after a few seconds. Keeping it
>> running for a few hours and it crashes. I have currently limited the calls
>> per second to 200 and max call session to 600, but still there are calls
>> where opensips responds slow to SIP packets.
>>
>> Here are a few configurations that i am using:
>>
>> fork=yes
>> children=200
>>
>> loadmodule "mi_datagram.so"
>> modparam("mi_datagram", "socket_name", "/tmp/opensips.sock")
>> modparam("mi_datagram", "socket_name", "udp:localhost:2000")
>> modparam("mi_datagram", "children_count", 200)
>>
>>
>> Do you have any pointers where i should start looking at? I have also
>> generated core dump files let me know if you need some more info on this.
>>
>> Regards,
>> Qasim
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20170405/0300f7c2/attachment.html>