[Kamailio-Users] Retransmitted SUBSCRIBEs

[catalina oancea]

Sorry for so many messages. I was using the sipp simulator which was
sending different branch-es in the retransmission. The bug occured
with Snom because I didn't have t_newtran in my cfg, and then
replicated again with t_newtran because I was using sipp to test.
So actually t_newtran did the trick.

Thanks





2009/10/9 Klaus Darilion <klaus.mailinglists at pernau.at>:
> Try xlog() after t_newtran. In case of the retransmission, you should not
> see the xlog as it should be absorbed silently by t_newtran.
>
> regards
> klaus
>
> catalina oancea schrieb:
>>
>> Hi
>>
>> Thanks for answering. I tried with t_newtran. It still doesn't work.
>> The error is:
>>
>> ERROR:presence:get_stored_info: wrong sequence number;received: 2 -
>> stored: 2
>>
>> Thanks
>> Catalina
>>
>>
>> 2009/10/9 Klaus Darilion <klaus.mailinglists at pernau.at>:
>>>
>>> catalina oancea schrieb:
>>>>
>>>> Hello,
>>>>
>>>> If a phone re-sends a presence SUBSCRIBE requests even after 5 or 500
>>>> milliseconds, kamailio replied with '400 Bad request' to the second
>>>> SUBSCRIBE
>>>>
>>>> The scenario is:
>>>>
>>>> Phone                           Kamailio
>>>>  SUBSCRIBE    ---------->
>>>>        401    <----------
>>>>  SUBSCRIBE    ---------->
>>>>        202    <----------
>>>>      Pause    [      5ms]              (or 500 ms, the same thing
>>>> happens)
>>>>  SUBSCRIBE    ---------->         (previous SUBSCRIBE retransmitted)
>>>> 400 Bad request <----------
>>>>
>>>> Is this the expected behavior? Is there any way to get kamailio to
>>>> recognize the retransmission?
>>>> This causes problems with Snom phones because they receive 400 reply
>>>> when they retransmit the SUBSCRIBE, and presence stops working.
>>>
>>> First, it would be a good idea to handle the SUBSCRIBE stateful, I think
>>> his
>>> can be done by calling t_newtran() before, and t_release() after
>>> handle_subscribe().
>>>
>>> regards
>>> klaus
>>>
>