Hi Daniel,

Thar is correct, the top Via header branch param differs (has a .1 instead of .0 at the end), that doesn't seem to count according to RFC 3271 section 8.2.2.2

Looking at the sip trace, the ACK for 415 is sent out before the 2nd branch INVITE.


Unfortunately, inserting a 2 seconds sleep didn't help :(

Regards,
--Sergiu


On Mon, May 7, 2018 at 6:21 AM, Daniel-Constantin Mierla <miconda@gmail.com> wrote:

Hello,

the branch parameter value in top Via header is different, so this should indicate is not the same branch.

Have you look at the network traffic, is the ACK for 415 getting first to device, before the 2nd branch INVITE? A that moment, the transaction should be considered terminated and a new one started.

My understanding of "merging request" is that it should be done if two requests arrive at the same time, both still being active, without a final response sent back.

What you can try to see what happens is to add a sleep() of few seconds for the 2nd branch and see if this time the device cleared the previous transaction. Of course, not like a really good solution, but can reveal what the device is doing.

Cheers,
Daniel


On 04.05.18 20:43, Sergiu Pojoga wrote:
In continuation of previous thread about none/optional/compulsory SRTP handling...

I used the suggested approach of t_reuse_branch() in a tm:branch-failure detected by t_check_status("(415)|(488)") replies, with the only difference of me doing it vice-versa from the suggested mantra of offer SRTP first then fallback to RTP.

It works... well, for some UACs, not so with others.

The 'good' UACs happily accept an updated branch, like for e.g. Yealink, will reply with "488 Not Acceptable Here" at first followed by a 200 OK.

The 'bad' UACs, like for e.g. Bria || Zoiper, will issue a "415 Unsupported Media Type" at first, and then a "482 Merged Request" after receiving and updated branch with a different RTP profile.

Rightfully so, they decided that the updated branch is a retransmission, since both branches have the same CSeq, Call-ID and From tag. 
The branch ID however differs, so does the RTP profile obviously.

Any suggestion on how to convince all UACs that the 2nd attempt is different than the 1st one?
According to RFC 3261,  a request is a merged request if ...

"8.2.2.2 Merged Requests

   If the request has no tag in the To header field, the UAS core MUST
   check the request against ongoing transactions.  If the From tag,
   Call-ID, and CSeq exactly match those associated with an ongoing
   transaction, but the request does not match that transaction (based
   on the matching rules in Section 17.2.3), the UAS core SHOULD
   generate a 482 (Loop Detected) response and pass it to the server
   transaction."



_______________________________________________
Kamailio (SER) - Users Mailing List
sr-users@lists.kamailio.org
https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users

-- 
Daniel-Constantin Mierla
www.twitter.com/miconda -- www.linkedin.com/in/miconda
Kamailio World Conference - May 14-16, 2018 - www.kamailioworld.com