[OpenSIPS-Users] solved => Re: explizit handling auf replyto
Uwe Kastens
kiste at kiste.org
Fri Sep 25 17:23:37 CEST 2009
Hi Bogdan,
Thanks again for your help.
I had a problem in my script, that causes some "loop" on one opensips
server. This causes, that the opensips server gets an INVITE from
itself, which was dropped by a security rule. After solving the cause
for that looping the expicit handlling - which caused some other trouble
- was not needed anymore.
BR
Uwe
Bogdan-Andrei Iancu schrieb:
> Hi Uwe,
>
> Sorry for the delay - I took a look at the logs and there is what I
> understand from there:
> - I guess you do a parallel forking as the 486 comes from a second
> branch (id 1 starting from 0) (see ;branch=z9hG4bK52ad.7d455a97.1 )
> - The TM decides to store the reply without relaying
> "DBG:tm:relay_reply: branch=1, save=1, relay=-1" TM does not relay it
> as probably there is no final reply on the first branch (id = 0).
>
> So, the final reply is not forwarded because not all branches are
> completed yet, so TM cannot pick yet a final reply to be sent to UAC.
>
> It is a typical behaviour during parallel forking.
>
> Regards,
> Bogdan
>
>
> Uwe Kastens wrote:
>> Hello,
>>
>> Ok. I need to have that forward() in my configuration the get answers
>> like 404, 486, 487 back to my asterisk. Reading your statements this
>> should not be possible since I use t_relay for the requests and the
>> replys should be routed by default.
>>
>> I will make a trace and post it to the list. One with forward and the
>> other without.
>>
>> BR and Thanks in advance
>>
>> Uwe
>>
>>
>>
>> Bogdan-Andrei Iancu schrieb:
>>
>>> Uwe,
>>>
>>> forward() is a function exclusivly used for REQUESTS - for replies,
>>> nothing needs to be done as OpenSIPS will do it automatically:
>>>
>>> 1) if the requests was statefully forwarded (via t_relay() ), the
>>> transaction will contain all the info to route back the reply
>>>
>>> 2) if the requests was statelessly forwarded (via forward() ), the VIA
>>> stack (in received reply) will contain all the info to route back the reply
>>>
>>> Regards,
>>> Bogdan
>>>
>>>
>>> Uwe Kastens wrote:
>>>
>>>> Hi Bogdan,
>>>>
>>>>
>>>>
>>>>>> I need to route this
>>>>>> replys with an reply_route and forward them explicitly to the pstn gateway.
>>>>>>
>>>>>>
>>>>>>
>>>>> and how do you do reply routing ?? replies are automatically routed
>>>>> based on VIA stack and you cannot influence this from script.
>>>>>
>>>>>
>>>> ...
>>>> if (!t_relay()) {
>>>> sl_reply_error();
>>>> };
>>>>
>>>> t_on_reply("1");
>>>> }
>>>> onreply_route[1] {
>>>> xlog("L_DBG","========== route [1] t_on_failure msg-len from $si:$sp:
>>>> $ml \n$mb\n ==========\n");
>>>> forward("10.20.30.101:5100");
>>>> }
>>>>
>>>> BR
>>>>
>>>> Uwe
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>> Regards,
>>>>> Bogdan
>>>>>
>>>>>
>>>>>> This not as it should be?
>>>>>>
>>>>>>
>>>>>>
>>>>>> BR
>>>>>>
>>>>>> Uwe
>>>>>>
>>>>>>
>>>>>>
>>>>>>> using forward() forces a stateless behaviour of OpenSIPS. To be able to
>>>>>>> catch replies for a requests (what on_reply route does) it requires a
>>>>>>> statefull approach. This is the reason it does not work for you.
>>>>>>>
>>>>>>> If you use t_relay() instead of forward(), you should be able to use
>>>>>>> failure and onreply routes.
>>>>>>>
>>>>>>> Regards,
>>>>>>> Bogdan
>>>>>>>
>>>>>>> Uwe Kastens wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> Hello,
>>>>>>>>
>>>>>>>> I am using opensips 1.5.1 and I have the problem, that busy messages are
>>>>>>>> not passed to the mediagw. Anything else is working (ACK etc.pp.). If I
>>>>>>>> route that message via forward("IP:port") to the media gw in on_replyto,
>>>>>>>> busy is processed correctly. If not, the busy is not send to the mediagw.
>>>>>>>>
>>>>>>>> So I was wondering if I had to handle some replyto messages?
>>>>>>>>
>>>>>>>> BR
>>>>>>>>
>>>>>>>> uwe
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
--
kiste lat: 54.322684, lon: 10.13586
More information about the Users
mailing list