Hi Callum,

I would say you are on the right tracks.

For detecting different NAT situation, there is no other way than to play with the flags of the `nat_uac_test()`.....

For the in-dialog NAT handling, during the initial request you need to learn the NAT status of the parties involved (for the caller, based on the nat_uac_test(), for the callee, based on the data from lookup()).  Once you learned the caller/callee 's NAT status, you should preserve this information across the whole dialog, so, when handling in-dialog requests, you already know the NAT status of each party.

And yes, when you receive traffic (requests or replies) for a party behind the NAT, you need to do the Contact fixing (and SDP if the case), in order to get rid of the private IPs.

Best regards,

Bogdan-Andrei Iancu

OpenSIPS Founder and Developer
  https://www.opensips-solutions.com
OpenSIPS Summit, Amsterdam, May 2020
  https://www.opensips.org/events/Summit-2020Amsterdam/
OpenSIPS Bootcamp, Miami, March 2020
  https://opensips.org/training/OpenSIPS_Bootcamp_2020/

On 12/13/19 2:12 PM, Callum Guy wrote:
Hi All,

I am operating a registrar which proxies calls to an internal network
of media servers.

Most of my subscribers are operating using RFC1918 addresses behind
NAT. We detect this configuration through nat_uac_test() and patch the
SIP using fix_nated_contact(). By rewriting the requests the media
servers will send in-dialog requests with request URI set to the NAT
address.

This works fine for the majority of my use cases however I am now
dealing with a UAC which has chosen to use public addresses on the
handsets but continues to run behind NAT. The effect here is that my
choice of flags for nat_uac_test() do not match the scenario and the
rewrites are not happening.

I can resolve this issue with some additional flags and logic however
I wonder if there is a "more correct" way to do this. In an ideal
world I would lookup the registration during INVITE processing, notice
that there is a different received address and use this for all future
communications for the rest of the dialog.

For example the initial requests out to these handsets are performing
a lookup() operation which will retrieve the original contact for use
as request URI and received address for use as destination URI
allowing the request to be properly formed and forwarded. The issue
arises when the dialog is started, the UAC responds to the well formed
INVITE with 200 OK and unless I patch the contact with the received IP
the subsequent ACK will end up routing back to the UAC contact rather
than the NAT device. What I am hoping is that there may be a correct
way for me to track the received IP against the dialog such that it
can be used in subsequent in-dialog requests, allowing the request URI
to correctly represent the UAC contact whilst still delivering
requests to the NAT address. I can probably achieve this by recording
the information using dialog variables however I imagine this is a
common scenario so I wouldn't want to add this logic in manually if
there was a proper way to do this natively in OpenSIPs.

Hopefully this isn't one of those questions that gets asked too
regularly however if it is please point me in the direction of an
article that might help, I've re-read the nathelper, nat_trasveral,
registrar and usrloc documentation today and haven't found what I'm
looking for. Any pointers would be appreciated before I embark on a
homegrown solution.

Thanks,

Callum



_______________________________________________
Users mailing list
[email protected]
http://lists.opensips.org/cgi-bin/mailman/listinfo/users

Reply via email to