[sr-dev] force_rtp_proxy() usage with IP that is not natively homed

[Alex Balashov]

Sorry for cross-posting from k-users; was hoping maybe someone here 
has some insights, as rtpproxy is a sparsely documented topic.

-------- Original Message --------
Subject: force_rtp_proxy() usage with IP that is not natively homed
Date: Mon, 02 Nov 2009 01:23:40 -0500
From: Alex Balashov <abalashov at evaristesys.com>
To: kamailio <users at lists.kamailio.org>

Greetings,

Please excuse the apparent silliness of this request, but I am curious
  if there is a combination of flags passed to force_rtp_proxy()
and/or rtpproxy invocation options that can allow force_rtp_proxy() to
substitute an SDP endpoint IP that is not a physical device/interface
on the host on which rtpproxy is running?

Let me explain further:

1. I have a host with a single network interface that has private IP
192.168.1.254.

2. I am receiving an inbound call from outside the network, coming in
on a public IP address 4.4.4.4 that is a 1-to-1 DNAT/SNAT to
192.168.1.254.

3. The call is forwarded to a PBX on the same LAN as the native
interface, e.g. 192.168.1.250.

4. I would like to set up SDP in such a way that 192.168.1.250
sendsmedia to the 192.168.1.254 interface while the outside host
initiating the call sends its media to 4.4.4.4.


Under normal circumstances in which both the public and private
interface were physically present on the machine, this would call for
the use of "bridging" mode (rtpproxy invoked with -l
4.4.4.4/192.168.1.254 and use of "ie" and "ei" flags with
force_rtp_proxy() - we have previously established that
rtpproxy_offer/answer do not actually seem to work as advertised).

But is it possible to "bridge" to an IP that is not routed to the box?

I expected that force_rtp_proxy()'s second parameter - ip_address -
would do this.  However, this does not appear to let me override the
SDP endpoint with any IP address I want;  it seems that I am still
limited to choosing from those IPs physically available on the
rtpproxy host.  I would guess that this is because ip_address is
passed as a parameter through rtpproxy's control API socket and is
validated by rtpproxy itself, and if it is not a local existing
address, is overridden with one that is.

Is there a clever combination of flags that can be used to accomplish
this, or is it a lost cause?

Thanks!

-- 
Alex Balashov - Principal
Evariste Systems
Web     : http://www.evaristesys.com/
Tel     : (+1) (678) 954-0670
Direct  : (+1) (678) 954-0671


-- 
Alex Balashov - Principal
Evariste Systems
Web     : http://www.evaristesys.com/
Tel     : (+1) (678) 954-0670
Direct  : (+1) (678) 954-0671