[Serusers] Please HELP: TM Module Question - sr-users

22 Mar 2005

Hi All.

I have a problem which I think is a Sonus issue, but I need to make
sure before I blame someone else for what could be my problem.

The issue is related to re-INVITE messages from a Sonus PSTN gateway
to my SER-0.9.1 sip proxy.

Sometimes the Sonus box sends us a retransmitted re-INVITE for which
SER just sent a 200OK.  It occasionally does this even considering we
originally reply with a 100Try message.

So what SER then does is forwards the retransmitted re-INVITE to the
SIP UA, but the SIP UA ignores this re-INVITE because it just sent a
200OK for the original re-INVITE. SER keeps trying to send the
re-INVITE to the SIP UA until it times out, then it replies to the
Sonus GW with a 407 Request Timeout and then Sonus disconnects the
call and sends SER a BYE message.

Now what I didn't mention yet is that when ser sends a 200OK response
back to the Sonus box in reponse to the original re-INVITE, rather
than getting an ACK back from Sonus we get a retransmitted re-INVITE
and then we get the ACK.

So since the SIP UA is smart enough to know not to response to the
retransmitted re-INVITE, shouldn't SER also be smart enough and just
absorb it rather than relaying it to the SIP UA?

Anyhow a picture is worth a million words so here is a SIP message log
that begins with a re-INVITE from the Sonus GW.

NOTE 1: SER is located behind a Cisco 3600 so the ALG functionality in
the 3600 has rewritten all the IP address (ie, the 10.3.0.0 network).

NOTE 2: This problem seems to be random. Some calls are very short and
others are very long before the disconnect happens - and still others
are perfectly fine for the call duration.

NOTE 3: Sonus sends SER a re-INVITE every two minutes.

Can anyone please confirm that ser has indeed handled this re-INVITE
properly and that Sonus is actually incorrectly processing calls?

Is there a way to hack a band-aide here in the event that the problem
is Sonus and they will not fix their side?

Regards,
Paul