[SIPForum-techwg] Could to-tag in 200 OK response be different with the to-tag in 180 ringing?
Dale Worley
dworley at pingtel.com
Fri Jul 27 09:52:16 EDT 2007
On Fri, 2007-07-27 at 12:49 +0800, mike xu wrote:
> Do you know that in RFC3261, could the to-tag in 200 OK be different
> with the one in 180 ringing?
It is allowed, but it does not mean what you might expect. Each to-tag
represents one dialog. The 180 has a to-tag value, and it is part of a
dialog. The 200 has a different to-tag value, and so it is part of a
*different* dialog. The 180's dialog never succeeds (because the dialog
never returns a 200 response), and eventually the UAC deletes knowledge
of it. So from the UAC's point of view, this is an example of forking.
(In general, a UAC must be able to track multiple early dialogs until it
receives a 200. After that, it may also receive 200 responses for other
forks of the INVITE, which it must either handle or terminate by sending
BYE.
> Below is the case which I met, in step F1 (180 ringing), callee
> generated one to-tag (46bf8) while in step F2 (200 OK) the sip server
> generates a new to-tag which is "as5123fc8f", and my sip endpoint
> didn't match this 200 OK response, the sip server will resend 200 OK
> and I obveriously couldn't hear the voice mail info.
>
> Caller SIP_SERVER Callee
> | | |
> | INVITE | |
> |------------->| INVITE |
> | (100 Trying) |----------->|
> |<-------------| |
> | |180 Ringing F1 |
> | 180 Ringing |<-----------|
> |<-------------| |
> | | CANCEL |
> | |----------->|
> | | 200 OK |
> | |<-----------|
> | |487 Req Cancel |
> | |<------------|
> | | ACK |
> | |------------>|
> |200 OK(SIP/SDP) F2| |
> |<-------------| |
> | ACK | |
> |------------->| |
> | voicemail RTP | |
> |<===============|| |
>
> Does any body know that the new to-tag in F2's 200 OK response is
> valid or invalid? Does it comply to RFC3261?
It looks like SIP_SERVER is doing what it should -- the 200 is
establishing a different dialog than the one that the 180 was reporting
on.
Dale
More information about the techwg
mailing list