[OpenSIPS-Users] mediaproxy relay stuck on source port 1024

[Jeff Pyle]

Hello,

I have Mediaproxy 2.6.1 and OpenSIPS 1.10 on Debian wheezy with a bit of an
unusual configuration.

The box has a LAN-facing interface ('lan') and a WAN-facing interface
('wan').  The lan has a private IP and the wan has a public IP.  Each
interface has a separate OpenSIPS instance bound to it for communication
with SIP entities on its respective network.  The two OpenSIPS instances
exchange SIP traffic with each other across localhost.

I had rtpproxy bridging media across the interfaces but it has a problem I
can't work around at the moment.  So, I'm attempting to creatively
configure an RTP bridge with Mediaproxy.  I've done this by running a
separate relay/dispatcher pair on each OpenSIPS instance.  Each pair has
its own config with the correct lan or wan IP address.  Since two
interfaces on a single linux system can send traffic directly to each other
without concern for subnets and traditional routing rules, I figure if I
can get each relay to send traffic to the other one and then out thenI've
created a bridge.

The cool thing is it (almost) works.  On the relays I've configured
different port ranges to make it more obvious which relay I'm observing,
with 16384:24572 on the wan instance and 24576:32768 on the lan instance.
All the SDPs look good through the system with proper port selection.  The
relay debugs and dispatcher stats agree with the SDPs.  RTP flows look just
about right, with the exception that traffic leaving the relay always has
source port 1024.

Any thoughts on what might cause a relay's traffic always to source from
1024 rather than the port it has chosen?  This is the case for both the lan
and wan relays sending traffic to its respective host on its own network.



Regards,
Jeff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20140924/0bd217e9/attachment-0001.htm>