[Users] Locally generated single-hop ACKs with Route headers

Klaus Darilion klaus.mailinglists at pernau.at
Wed Jul 26 13:49:17 CEST 2006


Augustin, Andrew (external) wrote:
> Hello,
> 
> Hi Klaus, Thanks for your response. In answer to your question: yes,
> the INVITE does contain a Route header.
> 
> 
> Further info on the problem:
> 
> There are 3 proxies between A and B parties.
> 
> The INVITE message is routed from A-Party through Proxy-1, through
> Proxy-2 before finally reaching B-Party. The Route header first
> appears in the INVITE message sent by Proxy-1 and is removed at
> Proxy-2.

Thus, the Route header between P1 and P2 indicates P2 as next hop, correct?

> There is an errored response message to this INVITE (e.g. 'Busy' or
> 'Request Terminated'). The errored response is acknowledged with a
> one-hop locally generated ACK at each point between B and A.
> 
> I expected the ACK at Proxy-2 to take the same one-hop route as the
> original INVITE at Proxy-2, i.e. the ACK sent by Proxy-2 will contain
> routing data based on routing data from the INVITE previously sent by
> Proxy-2.
> 
> What actually occurs is that the Route header in the ACK sent by
> Proxy-2 is a duplicate of the Route header of the INVITE that Proxy-2
> has RECEIVED from Proxy-1 and NOT that of the INVITE message it had
> previously sent.
> 
> Are you aware of this problem and any resolution and/or workarounds?

Hi, I'm not aware of this problem. But it sounds like a bug in openser 
when generating the ACK. I do not know if it possible to have a 
workaround in openser.cfg, but anyway if this is a bug it should be 
fixed in ACK generation, not in the routing logic.

Please make a bug report at the sourceforge bug tracker, maybe Bogdan or 
Daniel can take a look at it.

Regards
Klaus

> 
> 
> We would like to remove this unwanted Route header from the ACK. How
> is it possible to do this (i.e. which routing block?)?
> 
> The known issue described in the previous email (below) appears to
> prevent us from removing this header from the ACK. Are you or anyone
> else in the OpenSER community aware of any fix or resolution to this
> problem?
> 
> 
> Best regards, Andrew Augustin
> 
> -----Ursprüngliche Nachricht----- Von: Klaus Darilion
> [mailto:klaus.mailinglists at pernau.at] Gesendet: Dienstag, 25. Juli
> 2006 18:15 An: Augustin, Andrew (external) Cc: users at openser.org 
> Betreff: Re: [Users] Locally generated single-hop ACKs with Route
> headers
> 
> 
> I wonder why there are route headers in the ACK - they are only 
> necessary if the INVITE also hed route headers (RFC3261, 17.1.1.3)
> 
> regards klaus
> 
> Augustin, Andrew (external) wrote:
>> Hello,
>> 
>> I am using OpenSER v1.0.1
>> 
>> I am currently testing with proxies running OpenSER.
>> 
>> I have a problem with ACK messages generated and routed for
>> scenarios where an inititiated call session fails to establish
>> between A and B parties e.g. Unsuccessful No Answer, Unsuccessful
>> Busy. The ACK messages are single hop locally generated responses
>> to the sip error message e.g. '487 Request Terminated', '486 Busy'.
>> 
>> 
>> The behaviour I observe is that Route headers are 'remembered' from
>>  the initiating INVITE message and inserted into the ACK messages 
>> automatically generated by OpenSER.
>> 
>> This extra Route header causes problems with routing and the 
>> generation unwanted messages (error message resends) in addition to
>>  expeceted scenario call-flow. It is also visible at the B-Party.
>> 
>> I would like to either prevent the creation of, or remove this
>> Route header. I have tried to remove this header using the OpenSER
>> available methods, but this has failed.
>> 
>> I have noticed that the known issue described in the document: SIP 
>> Express Router v0.8.8 - Developer's Guide under section: 9.6.4
>> Known Issues appears to describe this problem.
>> 
>> "local ACK/CANCELs copy'n'pastes Route and ignores deleted Routes"
>> 
>> Has this issue been resolved? If so, from what version of OpenSER?
>> 
>> If it has not been resolved, is there a way to resolve or
>> workaround this problem?
>> 
>> Best regards, Andrew Augustin
>> 
>> 
>> ----------------------------------------------------------------------
>>  --
>> 
>> _______________________________________________ Users mailing list 
>> Users at openser.org http://openser.org/cgi-bin/mailman/listinfo/users




More information about the Users mailing list