>>> "Matthew Kitchin (public/usenet)" <[email protected]>only        
>>> route, or does it NAT?  I would suspect its actually NATing a        
>>> 172.x.x.x IP to the 10.83 side if your changing the port.
      </[email protected]><[email protected]> 01/27/11 2:02 PM 
>>>
            >>On 1/27/2011 12:53 PM, Matt White wrote:        It is not NATing 
the IP. Their subnet has direct (routed) access to    ours. The router sits 
between these 2 subnets and in this case only    translates the port. Not the 
IP. It is a Verizon DS3, and we have    100 MPLS sites coming in >>through it. 
Also, in case it is relevant, I    have to put 'some' info on the NAT setting 
under servers in the sipx    gui. I don;t use STUN, so I change it to Specify 
an IP and I have to    put something in. It was trial and error 1.5 years ago 
when we    started this. I put in the IP for the >>Verizon gateway we send our  
  calls. I know that isn't correct, but I have to put some value in.    I'm not 
sure if that is relevant. The only other setting we change    is 'Use public 
address for        call setup' under Gateway, ITSP account. We have to uncheck 
that        box. 

Interesting.  I'm surely not a CCIE, but port translation always requires 
natting at least on a cisco.  Even if your just trying to change the outbound 
port it requires the "ip nat" command.

If under "servers"..."NAT" and then public IP address you have put the ip 
address of the verizon sip proxy...that would certainly be wrong.  This needs 
to be the IP address that sipxbridge receives sip on...whether its NATTED 
public ip or its private ip...either way its the ip that sipx will get traffic 
on and tell others where to send traffic destined for itself.

Change from port 5060 to 5080 using PAT/NAT does work.  We have many many 
installs using this method for carriers that must send to port 5060.  

But I suspect your setup is stuck somewhere between thinking its natted and not 
being natted.  You should pick one and commit to it.

The trace shows the phone send directly to sipxbridge and it shouldn't.  When 
the codec doesn't change it looks like sipxbridge balks becuase it didn't come 
form the proxy, but then accepts it because the RTP stream is already open.  
This doesn't work when the codec has to change.  But if the real issue isn't 
fixed...this will always be an unreliable setup because your breaking the rules 
of sipx by sending around the sipxproxy.

-M
  </[email protected]>
_______________________________________________
sipx-users mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users/

Reply via email to