2011/5/10 Bob Penfield <[email protected]>:
> In RFC 3261, section 16.7 (Response Processing for proxies) on page 110, it 
> says:
>
>         After a final response has been sent on the server transaction,
>         the following responses MUST be forwarded immediately:
>
>         -  Any 2xx response to an INVITE request
>
> Thus the proxy must forward the 2xx when the client transaction is in the 
> "completed" state.

Hi Bob, I meant that the 2xx arrives with same Via branch as a 4xx
received before. So in this case there is just a single client
transaction in the proxy so text above is not longer valid.

This is:

1) Proxy creates a client transaction and receives a 4XX => completed state.
2) Later a 2XX arrives with same branch (so it's matched by the same
client transaction as there is no more client transactions).
3) IMHO it's clear that it should be ignored. Again: it's not a 2xx
received within a different client transaction.

Point 2 should never occur, but I'm in the case that a broken UAS or
downstream proxy forwards such 2XX response.


> If the client transaction has already been deleted, then this text (on page 
> 107) applies:



>   When a response is received by an element, it first tries to locate a
>   client transaction (Section 17.1.3) matching the response.  If none
>   is found, the element MUST process the response (even if it is an
>   informational response) as a stateless proxy (described below).  If a
>   match is found, the response is handed to the client transaction.
>
>      Forwarding responses for which a client transaction (or more
>      generally any knowledge of having sent an associated request) is
>      not found improves robustness.  In particular, it ensures that
>      "late" 2xx responses to INVITE requests are forwarded properly.
>
> Therefore, the proxy always forward all 2xx responses to INVITE.

This is not true after RFC 6026 ;)


-- 
Iñaki Baz Castillo
<[email protected]>

_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors

Reply via email to