>Now SIP phone A dies and never comes back.
>Furthermore, nobody is  
>available to answer C, so it keeps ringing, and
>ringing, and  ringing......


Typically UAS (of C) shall start a timer on receiving
INVITE. if C user does not answer the call within that
timer, then UAS shall send back appropriate error
response and shall stop Ringing. On receiving this
error response, UAC shall terminate the call. This
error response shall also terminate the transaction of
any stateful proxy(Proxy B in your case) in the path.

The UAC also starts a timer after initiating the call
with INVITE. If that timer expires before receiving
200 OK then it shall send CANCEL to terminate the
INVITE transaction. In fact, 487 response of INVITE
(followed by INVITE) terminates the INVITE dialog.

Therefore, depending on which timer expires first,
either A shall send CANCEL to cancel INVITE or C shall
send error response to terminate INVITE dialog in your
case.

>Now if SIP Phone A comes back up and calls C again
>(through B), we will get a 486 busy.

If UAC and UAS starts the timer as specified above,
SIP phone shall not get 486 Busy in your described
scenario.


>The question is when will the 486 busy clear up?

Not sure whether your are asking about UAC or UAS. The
486 shall definitely clear up the dialog at UAC
initiated by INVITE. But 486 of new INVITE shall not
typically impact any exiting orphaned dialog.


>Is the proxy and C suppose to clear up after a time 
> and cancel the call?    (What if A  
>called C directly.... since A died, who would CANCEL
>the call on C?)

If the proxy is the transaction stateful proxy then it
must start a timer so that it can clean-up its
transaction in case network failure(on the expiry).




                
__________________________________________________________
Yahoo! India Answers: Share what you know. Learn something new
http://in.answers.yahoo.com/
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors

Reply via email to