Hi Iñaki, Thanks a lot for the clarification - I will keep in mind this.
So most important implication of this is to remember and not to use the via versus received port test (as NAT test) for the TCP proto... Thanks and regards, Bogdan Iñaki Baz Castillo wrote: > 2009/2/25 Bogdan-Andrei Iancu <[email protected]>: > >> Hi Iñaki, >> >> to be honest, I'm note 100% sure about this. If you consider that the reply >> has to be routed back via the same TCP connection, what you have in VIA is >> the origin of the TCP conn (used by client to open the connection) and not >> the listening info (for receiving new connections). >> > > Bogdan, AFAIK what you say is not correct. I show you an example of a > TCP request sent by Twinkle: > > T 10.10.10.120:33947 -> 10.10.10.121:5070 [AP] > INVITE sip:[email protected] SIP/2.0 > Via: SIP/2.0/TCP 10.10.10.120:5060;rport;branch=z9hG4bKsbnnahng > > As you see, the Via sent-by is the *listening* address of Twinkle. > > RFC 3261 states more or less clearly that, in case of a TCP > connection, the response is sent using the existing TCP connection. > But the Via remains being the listening address. > In case the response cannot arrive due to a TCP error, the UAS should > open a *new* connection against the Via sent-by address (in the above > example it would be "10.10.10.120:5060", this is, the address in which > the UAC is listening). > > I'm 101% sure of it :) > > > > >> In such a case, the port should match...but again, I'm not 100% what you >> should put in VIA when doing TCP - the desctiption of the TCP connection or >> the description of the listening interface. >> > > The listening interface, sure :) > > > > _______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
