Summary:	ASTERISK-12809: Occassional failure to connect pickup call with park call
Reporter:	Alistair Cunningham (acunningham)	Labels:
Date Opened:	2008-10-01 23:18:21	Date Closed:	2009-01-22 14:45:45.000-0600
Priority:	Minor	Regression?	No
Status:	Closed/Complete	Components:	Core/PBX
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:	( 0) park1.txt
Description:	On a test machine, around 10% of the time, a call to 701 from an AGI() to pick up a call parked there fails to connect the two calls. Asterisk reports that the picking up call has been answered, but the parked caller continues to hear music, and the picking up caller hears silence for 10 seconds before the call times out. The call flow on such calls is identical to the other 90% of calls which work, so there seems to be a random component to this. ****** ADDITIONAL INFORMATION ****** This machine has the following in features.conf: [general] context => parkedcalls parkext => *7 parkingtime => 300 parkpos => 701-799
Comments:	By: Alistair Cunningham (acunningham) 2008-10-01 23:21:34 I have uploaded park1.txt, copied from Asterisk's full log, which shows where Asterisk tries to connect the calls together. A full trace of the entire call is available on request. Notice these lines: [Oct  1 17:28:37] DEBUG[4984] channel.c: Internal timing is enabled (option_internal_timing=1048576 chan->timingfd=27) There are a lot more of them (perhaps one for each RTP packet?) during the time the picking up call is hearing silence. By: Alistair Cunningham (acunningham) 2008-10-08 22:58:11 I've just been informed that this still occurs on Asterisk 1.4.22. By: Jeff Peeler (jpeeler) 2008-12-01 17:02:28.000-0600 Any additional information you can give would be appreciated. Can you try simplifying your setup a bit and see if the problem ever occurs? Local tests here with the simplest possible scenario of parking an extension and then dialing that extension from AGI has never failed. By: Alistair Cunningham (acunningham) 2008-12-01 17:28:30.000-0600 I'm afraid the AGI is Enswitch's enswitch_routed, which is over 6000 lines of code, so simplifying it is not really an option. We have only had one report of this on a test machine, so if it can't be fixed as is I suggest closing this ticket. By: Jeff Peeler (jpeeler) 2009-01-22 14:45:45.000-0600 Closed at the request of the reporter. Please reopen if you come across the problem again and are able to include the AGI causing the issue.