Hi,
first of all, excuse me for posting these questions twice, but the previous mail was bloated with too verbose sentences.
I'll be straightforward then.
1) Why do all the configuration files I was given to see on the web call record_route() when the method is *not* a REGISTER ?
2) What is the harm of calling fix_nated_contact() in addition to fix_nated_register() in a REGISTER message ? The configuration file [1] carefully avoids doing this.
[1] http://voip-info.org/wiki/view/OpenSER+And+RTPProxy
Thank you very much. Regards,
On 8/17/06, Jeremie Le Hen jeremie@le-hen.org wrote:
Hi,
first of all, excuse me for posting these questions twice, but the previous mail was bloated with too verbose sentences.
I'll be straightforward then.
- Why do all the configuration files I was given to see on the
web call record_route() when the method is *not* a REGISTER ?
You should understand what record routing is good for: inserting a sip node (in case of *ser it inserts itself) into the path that subsequent requests of an establishing session will follow. REGISTER is one of the sip's methods which does not establish any session (just like MESSAGE, for example). record_route()'ing REGISTERs won't do any harm, except wasting processor cycles.
2) What is the harm of calling fix_nated_contact() in addition
to fix_nated_register() in a REGISTER message ? The configuration file [1] carefully avoids doing this.
The harm is the same as above: if you do fix_nated_contact(), an AVP (with the originating address of the REGISTER, the one that SER sees in the networking layer, not the SIP layer) is set, instructing the registrar to disregard the body of Contact HF and consider only this AVP as true Contact.
[1] http://voip-info.org/wiki/view/OpenSER+And+RTPProxy
Thank you very much. Regards, -- Jeremie Le Hen < jeremie at le-hen dot org >< ttz at chchile dot org >
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Weiter Leiter wrote:
2) What is the harm of calling fix_nated_contact() in addition to fix_nated_register() in a REGISTER message ? The configuration file [1] carefully avoids doing this.
The harm is the same as above: if you do fix_nated_contact(), an AVP (with the originating address of the REGISTER, the one that SER sees in the networking layer, not the SIP layer) is set, instructing the registrar to disregard the body of Contact HF and consider only this AVP as true Contact.
More importantly, fix_nated_contact() modifies the URI of the Contact-HF of the 200-OK, whereas fix_nated_register() only appends parameters to the URI.
I'm not entirely sure if modifying the Contact URI of responses to REGISTERs violates RFC3261, because I don't know exactly how to interpret paragraph 10.3, number 8, but there are at least UACs which don't accept a 200-OK with modified Contact-URI.
Andy