[Users] call drop in a few seconds

unplug maillisting at gmail.com
Tue Feb 7 03:54:27 CET 2006


For the case below.

UA1 -----INVITE ----> UA2
UA1 ------call established---- UA2
UA1 ------BYE --------> UA2
UA1 ------INVITE ------> UA3  (in a few seconds)
UA1 ------call established---- UA3
UA1 <-------BYE -----------UA2
UA1 -----call drop--------UA3

It will cause call drop as the case in the previous discussion.  Does
it the same reason for the call drop?

Do you mean there is no way for openser to handle the case as it is
only a proxy?  Instead, UA1 device should handle the case such that it
should wait for the cycle completion before it really BYE/CANCEL the
call.  Am I right?

In my configuration, UA1 is a hardware sip phone and connect to the
openser server directly.

UA1 --------openser -----------CISCO gateway -------PSTN
(sip phone)

Do you mean the sip phone should handle the matching before it makes
another call?


On 2/3/06, Daniel-Constantin Mierla <daniel at voice-system.ro> wrote:
> On 02/02/06 17:28, unplug wrote:
> > Thanks for explanation.  As a normal user behaviour, he will cancel
> > the call when he noted that he has dialed a wrong number.  Then he
> > will make another call for his desired destination immediately.
> > As you said, UA1 needs to wait until the message cycle complete.
> > However, user have no idea when the message cycle will complete.
> The user should not be the one to solve the problem. The UA1 device has
> to behave properly, to keep the canceled dialog until the cycle
> completes. Moreover, it should match the BYE to the proper dialog or
> send back "call/leg transaction does not exist" when the call is no
> longer active.
> >  For
> > a long distance call, the time needed for completing the cycle will be
> > much longer.
> > Do you mean it can't be prevented in the voip/openser environment?
> >
> I would say no. In similar case to UA1 may be a gateway or multi-line
> phone which can have several calls in parallel. The UA1 must follow
> dialog matching specifications as per RFC3261 and everything should go
> just fine.
>
> Cheers,
> Daniel
>




More information about the sr-users mailing list