So it seems that I have to post my first question here. Till now I was able to solve every problem myself on setting up my first asterisk server, but now I’m quite clueless.
On incoming calls there is always an @ipaddress appended to the number of the calling party when signaled on the endpoint device. To rule out problems with the device itself I re-checked with a different one but same result.
This happens on internal calls and also on incoming calls via SIP trunk.
Strange thing is, when I use the IP address instead of domain name when registering the endpoint device the @ipaddress is omitted and working as I would expect it.
Placing an NoOp(${CALLERID(num)}) in the dialpan shows that it carries only the number so it seems to be appended elsewhere.
Any suggestions where and why @ipaddress is appended in that case?
We really need the full log with “pjsip set logger on” enabled, but I would point out that the domain part is not optional in what goes over the wire; it is the user part that is optional. Generally full logs should be run with verbosity set to, at least, 3.
The initial assumption is that the phone is not displaying the information in a friendly way.
Although the CALLERID function parameters look similar to what is in the From header, it hides the domain part, to make the caller ID look like a phone number
The from filed in the INVITE header shows “sip:username@ipaddress”, and that’s what the clients display (without protocol specifier) when a call arrived. It also seems that when the domain part of the from header matches the domain part of the receiving endpoint registration then it is omitted from displaying within the clients and only the CALLERID(num) is displayed.
To achieve what I expect as behaviour I found out that I could simply add the domain which is used for endpoint registration to the PJSIP endpoint template in the “from_domain” filed. In this case the IP in the INVITE header is replaced by the specified domain and the client omits the domain part as both are within the same domain.
The question remaining: Is this the expected behaviour or do I simply fix a misconfiguration in another place by adding the “from_domain” filed to the endpoint template.
Normally the registration for a local phone would use the domain name of the registrar, so you have an unusual configuration on the phone.
What the phone is doing seems to be a fairly sensible way of having the simple case behaviour look like a traditional phone, whilst reflecting that SIP uses URIs, not phone numbers.