[ previous ] [ next ] [ threads ]
 To :  paulc@v...
 From :  "martin A." <acevedoma@h...>
 Subject :  Re: [yate] about nat
 Date :  Mon, 19 Dec 2005 15:08:45 +0000
I have other question, If both endpoints are sip and they are behind a nat, 
can yate solve NAT? I have been trying this, but when yate sends INVITE to 
ep2, it sends to the private IP of ep2, What should I configure to resolve 
this?; or Is it not possible yet?, I have added full proxy and ignorevia=yes 
en ysip cofig file.... both End point have private addresses.....
thank you


>From: Paul Chitescu 
>To: yate@v...
>CC: "martin A." 
>Subject: Re: [yate] about nat
>Date: Sat, 17 Dec 2005 21:21:48 +0200 (EET)
>
>If the call comes on SIP you can decide on RTP forwarding per call. SIP
>always announces this capability and a decision on using it should be made
>during routing anyway.
>
>You can find on the site an example of how to process the rtp_forward
>parameter using the regexroute module. Creative use of sections and of the
>"call" (or "include" which is the same) statement can be used to enable
>RTP forwarding per user agent.
>
>For calls that arrive on H.323 the RTP forwarding decision is static as
>action has to be taken before most of call data (called number, aliases,
>endpoint type) is available. Currently this setting is global for the
>h323chan module.
>
>You MUST NOT deny RTP forwarding from routing if the call comes in using
>the H.323 protocol. Doing so WILL lead to trouble - either the call is
>refused or it will proceed but there will be no voice.
>
>Paul
>
>On Sat, 17 Dec 2005, martin A. wrote:
> > Hi, then you for your answer
> > [...]
> > Can I configure yate to forward rtp based on the name of the user 
>agent.. ?
> > I mean for each call...
> > I have some endpoints that are hardware(linksys) and we must use them in
> > proxy mode to traverse de NAT, but we can develope our own software 
>endpoint
> > for PC too, but I would like to forward the rtp for these endpoints in 
>order
> > to reduce the overhead of rtp proxy.... Is it possible?
> > [...]