[Freeswitch-dev] second leg of call isnt hanged up by Freeswitch

[kokoska rokoska]

Anthony Minessale napsal(a):
> yes do you have both a pcap and a console trace of the call.
> 
> start freeswitch with TPORT_LOG=1
> 
> TPORT_LOG=1 /usr/local/freeswitch/bin/freeswitch
> 
> set debug level
> 
>  > console loglevel debug
> 
> 
> capture all the text on the console
> 

I have just updated to current trunk and after a while I do the test and 
capture all possible dumps

> 
> my guess without seeing it is that user 22 is behind nat or a proxy or 
> something and you do not have keepalive turned on so the path for the 
> phone on 22 to send the bye to FS is gone so FS is not getting the bye 
> from phone 22
> 

No, no. There is nothing (except ethernet switch :-) between Freeswitch 
and UACs. They are all on the same lan subnet.

> We would never say 481 because if FS thinks the call leg should still be 
> up it will last until the call is terminated one way or another. 

But if one leg of call was hanged up (and "hangup after bridge" is set) 
I mean Freeswitch has to hang up second leg and than drop transaction.

> Remember, each leg of the bridged call is a separate sip call as we are 
> a b2bua.

Yes, I remember it :-)
But it IMO have nothing to do with 481. Every UAC or UAS has to be able 
to send 481 if transaction (call leg) doesn't exists. And from my 
oppinion most of UAs do it well (better to say, I have never heard about 
one didn't do it). It is mandatory by RFC3261, 21.4.19 and even Asterisk 
does it :-)


Thank you, Antohny, for your reply. I capture the packets and send them 
here...

Best regards,

kokoska.rokoska