I'm unclear on how multiple domains should be handled in sip/openser, etc. The scope of my confusion may be broader than just with openser, and this is more of a "concept" problem, but I'm hoping someone here can point me in the right direction.
I've read the "domain" module documentation at openser.org. I'm using a Linksys SPA942 IP phone for this experiment with my openser 1.0.1 server.
The SPA942 appears to do the following:
a) sends "sip:$PROXY" for the REGISTER request-uri. Here $PROXY is the actual entry typed in by the user, as opposed to something looked up by using DNS SRV.
b) appends "@$PROXY" to the "User ID:" entry in the "From:" field and appends @IP:PORT for "Contact:"
That is, as the person who configures the phone, I can view the proxy setting as declaring the domain of the user.
But it seems more natural to me to put the actual host name of the proxy in that setting. That is, if I run a service on sip.foo.net and have customers in domains bif.net and baz.net then each of them listing sip.foo.net as their sip proxy is reasonable.
But, if I go that route, then it seems the only way for the user to send the sip server their domain is as part of the userID: john@baz.net and if they do that then I get "From:" fields like john@baz.net@sip.foo.net, and similar "Contact:" settings.
Now, I can strip off the trailing domain but the clumsiness of that uri (sip:john@baz.net@sip.foo.net) gives me the feeling that this is the wrong approach. Looking at the "Contact:" pretty much convinces me of this: sip:john@baz.net@192.168.168.214:5060
But that takes me back to using the Proxy field as dual-duty: specifying the proxy and also the user's domain.
I just looked at a cisco 7905 and it too does not appear to have a way to set the user domain.
I can see how I could use unique AuthIDs to look up domains at registration time, but that seems misguided too.
What am I missing?
Thanks, -mark