Summary:ASTERISK-11688: SIP caller hanging up before answer does not stop Dial
Reporter:Gregor Schaffrath (kodomo)Labels:
Date Opened:2008-03-20 11:38:06Date Closed:2011-06-07 14:03:08
Versions:Frequency of
Environment:Attachments:( 0) debug
Description:If a SIP user hangs up before an answer, the Dial is not interrupted.
Instead, the following WARNING is issued:

[Mar 20 17:33:50] WARNING[20967]: app_dial.c:676 wait_for_answer: Unable to forward voice frame

and the receiver side continues ringing (thereby blocking the channel and causing costs, if the supposed call is picked up, eventually)

Comments:By: Mark Michelson (mmichelson) 2008-03-24 17:14:53

I have tried multiple times to reproduce this and was unable to. I noticed in your debug output that the destination phone operates under the chan_sirrix channel driver, which is unsupported. Could you test with a different destination channel type, such as SIP, and report if the same problem occurs? Thanks.

By: Francesco Romano (francesco_r) 2008-03-25 03:29:35

I have the same problem from revision 110618

UPDATE. I have reinserted in line 14658 of chan_sip.c this

if (p->owner)

and now works as expected. Perhaps mmichelson did an error in reverting the code.

By: Gregor Schaffrath (kodomo) 2008-03-25 07:33:20

...back from Easter vacations...

The problem seems to happen indeed only with external numbers. I've now contacted Sirrix support as well.
Maybe was there an additional API change w.r.t. calls hung up from the caller side, to which they need to adapt? (since I don't recall this problem happening before the update)

francesco_r - not sure, if it's the same error - for me, it doesn't happen with SIP/ or IAX2/ extensions... did it for you?

By: Mark Michelson (mmichelson) 2008-03-25 10:33:03


Yes, that section was not meant to be reverted. I'll fix it immediately.

By: Francesco Romano (francesco_r) 2008-03-25 10:38:13

My problem is only with the specific 110618 revision and only with SIP. However i will open another ticket because the behaviour is similar but the cause is totally  different.

UPDATE. i'll not open another issue...

By: Mark Michelson (mmichelson) 2008-03-25 10:40:44

francesco_r, no need, that section has already been reverted.

By: Gregor Schaffrath (kodomo) 2008-03-28 08:31:58

I did some more testing today and found something interesting (maybe I was wrong and it actually _is_ francesco_r's bug:
I tested three setups:
reroute: GXP2000<--SIP-->*1<--IAX2-->*2<--IAX2-->*1<--Srx-->ExternalPhone
(calls rerouted - i.e. looped - via a second system, currently not doing anything, but logging)
When cancelling the call, *2 didn't get a notice about that cancelling either.

direct: GXP2000<--SIP-->*1<--Srx-->ExternalPhone
Problem, as previously described.

private: GXP2000<--SIP-->*3<--IAX2-->*1<--Srx-->ExternalPhone
(calls placed via my own asterisk installation and routed over IAX2 to the system in question)
No problems... calls were being hung up instantly...
[I previously assumed that *3 and *1 have the same * version, which is not the case, so the fact that it worked there does not imply that it's not happening with out:IAX2 - sorry, my fault - was too tired when I tested it ;) ]

So it seems, as if the point, where things are turning bad, is really when the call is initiated via SIP and then is bridged to another channel type (which would indicate the problem to be located in chan_sip, as I initially suspected)

*1: Asterisk 1.4.18
*2: Asterisk
*3: Asterisk

By: Gregor Schaffrath (kodomo) 2008-04-01 09:56:27

putnopvut: Have you been able to reproduce it now? Is there something else, I could provide? As I wrote, it's actually happening in SIP->IAX2 cases, too, so I think it's a chan_sip problem after all...
francesco_r: has your patch been integrated into an official version, so I could try an upgrade, or could you provide a patch against the official 1.4.18?

By: Gregor Schaffrath (kodomo) 2008-05-21 09:07:35

UPDATE: upgraded to now available asterisk version 2.4.19
problem gone now (so it seems to have been the problem described by francesco_r)

By: Mark Michelson (mmichelson) 2008-05-21 09:11:09

Closing at reporter's request