[Serusers] fr_inv_timer + avp not working under high load

[Jiri Kuthan]

Hi Marcus,

I think we are now at the stage that the testing effort for Ottendorf is
more profittable than backporting effort. I understand that this is not
the answer you were hopeful to hear, but in the end of the day it is
about migration versus backporting evil and it seems that the latter
is more laborous for you.

-Jiri

At 14:17 06/11/2006, Marcus Hunger wrote:
>Hi Jiri,
>are you aware of a backport of the new timers from 0.10 
>to 0.9.6? Though we are going to test the prerelease, it is marked as 
>unstable, so it's hardly an option for us. Any help on this issue would be 
>great.
>
>Marcus
>
>On Monday 06 November 2006 10:17, Jiri Kuthan wrote:
>> At 05:40 06/11/2006, Nils Ohlmeier wrote:
>> >Hello Marcus.
>> >
>> >On Tuesday 31 October 2006 15:46, Marcus Hunger wrote:
>> >> Hi list,
>> >> I am trying to realise dynamic invite-timeout-timers loading
>> >> timer-settings per call from my database using avp. It seems to work
>> >> perfectly on my development-system but fails nearly completly in my high
>> >> loaded life-environment. According to my logs, timer values are set
>> >> correctly (eg.: reply_received: FR_INV_TIMER = 15, see
>> >> modules/tm/t_reply.c:1371), but most of the times the timers have no
>> >> effect.
>> >>
>> >> My setup is similar to
>> >> http://siprouter.onsip.org/doc/gettingstarted/ch09.html, but avps are
>> >> not written by hand, but acquired with:
>> >> modparam("tm", "fr_inv_timer_avp", "mytimer");
>> >> avp_db_load("$ruri/username","s:mytimer");
>> >>
>> >> 'mytimer' is an attribute in an usr_preferences-table.
>> >>
>> >> Ser's version is 0.9.6.
>> >>
>> >> I already searched the list-archives without finding anything regarding
>> >> this issue. I Also failed accessing the bug-tracker
>> >> (http://bugs.sip-router.org/), it gave me an internal server error.
>> >>
>> >> Has anybody experienced this effect? Is this a bug? Is there a
>> >> workaround?
>> >
>> >I'm not sure but the new timer implementation in SER head might solve this
>> >problem.
>>
>> I would definitely try this first. The thing is that the old timer system
>> was fast but not flexible, and variable timers just slowed down. The new
>> timer system should address that. Can you share with us if it made things
>> better for you?
>>
>> -jiri
>>
>>
>> --
>> Jiri Kuthan            http://iptel.org/~jiri/
>>
>> _______________________________________________
>> Serusers mailing list
>> Serusers at lists.iptel.org
>> http://lists.iptel.org/mailman/listinfo/serusers
>_______________________________________________
>Serusers mailing list
>Serusers at lists.iptel.org
>http://lists.iptel.org/mailman/listinfo/serusers

--
Jiri Kuthan            http://iptel.org/~jiri/