[OpenSIPS-Users] Parallel forking (using registrar branching) & mediaproxy

[erwan.humez at orange-ftgroup.com]

Hi,

I came across something weird about parallel forking: media information 
(IP @ received from SDP for RTP) is not properly communicated to media 
proxy, but only for one of the 2 calls.

Here is the context :
opensips 1.5.1 / mediaproxy 2.3.4 / Centos 5.2, on IP 192.168.2.233
Caller on sip softphone, on IP 192.168.2.223
Called on 2 tdm cisco media gateways, each registered using the same uri 
8003 at 192.168.2.233 behind 2 different IP@ : 10.0.0.2 and 10.0.1.1.
Objective is to achieve a call // forked from sip to both tdm called 
parties.

// sip forking seems to be working fine in the way that when one picks up 
the other one is properly cancelled. For information, both media gateways 
send a 183/SDP message and then a 200 OK/SDP upon pick-up. Media proxy 
sends/forwards 183/SDP messages to caller without modifying IP @ (keeping 
10.0.0.2 & 10.0.1.1) and then only sends the modified (with media proxy 
IP@) 200 OK/SDP message to the caller upon pick-up.

The issue (check below for media proxy traces) is reproductible because 
each time the same gateway is impacted (no need to precise that basic 
calls are working fine in both ways for call setup and on both gateways, 
which should prevent from any network related issue/question).

Of course i got all required traces that i can provide upon request to 
avoid spamming the list but so far and mainly, here is what i can see from 
media proxy traces:

Working call:

debug: Got traffic information for stream: (audio) 192.168.2.223:5298 
(RTP: 192.168.2.223:5298, RTCP: 192.168.2.223:5299) <-> 
192.168.2.233:50072 <-> 192.168.2.233:50074 <-> 10.0.0.2:19098 (RTP: 
10.0.0.2:19098, RTCP: 10.0.1.1:19163)

=> RTCP IP is not correct here but does not prevent the call to be 
successfully established.

Non-working call:

debug: Got traffic information for stream: (audio) 192.168.2.223:5304 
(RTP: 192.168.2.223:5304, RTCP: 192.168.2.223:5305) <-> 
192.168.2.233:50080 <-> 192.168.2.233:50082 <-> 10.0.1.1:16910 (RTP: 
10.0.0.2:19560, RTCP: 10.0.0.2:19561)

=> RTP 10.0.0.2:19560 information is taken from the 183/SDP message sent 
by the gateway to wich the forked call is canceled: it should be the RTP 
information coming from the 183/SDP message sent by the other gateway 
(10.0.0.1:16910).
=> neither RTP and RTCP are correct here. Network traces confirm that 
media proxy keeps sending RTP packets towards 10.0.0.2:19560 getting an 
icmp destination port unreachable of course.

I understand that SDP information received in both 183 messages could 
cause some troubles in opensips and media proxy communications... 
Somewhere and somehow, the information should be updated in media proxy...

... but maybe i missed or misunderstood something, so please can someone 
enlight me on this issue ?

Thanks a lot for help and support,

Erwan.

*********************************
This message and any attachments (the "message") are confidential and intended solely for the addressees. 
Any unauthorised use or dissemination is prohibited.
Messages are susceptible to alteration. 
France Telecom Group shall not be liable for the message if altered, changed or falsified.
If you are not the intended addressee of this message, please cancel it immediately and inform the sender.
********************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.opensips.org/pipermail/users/attachments/20090604/5ce5a1bd/attachment.htm