|Summary:||ASTERISK-08807: transfered call does not edn until timeout when closed from the transferee|
|Reporter:||Kaloyan Kovachev (knk)||Labels:|
|Date Opened:||2007-02-14 10:43:30.000-0600||Date Closed:||2007-02-16 12:15:25.000-0600|
|Environment:||Attachments:||( 0) full.atxfr.feature.tar.bz2|
( 1) full.blind.feature.tar.bz2
( 2) full.blind.refer.tar.bz2
( 3) pbx_config.diff
|Description:||A (1070 in the log) calls B (1601) then transfers the call to C (1602). The conversation is between B and C now. If B hangup first everithing is OK, but if C hangups (and B not) the call stais for few more sconds until the local timeout is reached. If during this timeout B also hangups then it ends immedeately.|
****** ADDITIONAL INFORMATION ******
attached are 3 full logs (error,warn,verbose,debug,dtmf logged where verbose and debug level 100, SIP debug enabled) with feature transfer - blind and attended and with SIP REFER transfer from the phone.
compiled with DEBUG_CHANNEL_LOGS, DEBUG_THREADS, DEBUG_THREADLOCALS, DETECT_DEADLOCKS, DO_CRASH and DONT_OPTIMIZE
The full.blind.feature one is when B ends the call and all is OK, while the other two are showing the problem.
|Comments:||By: Kaloyan Kovachev (knk) 2007-02-14 10:50:24.000-0600|
The subject should be "... does not end ..." instead of "... does not edn ...".
BTW There is still an error message about unlocking the channel more times than locked (full.blind.refer) as in bug (7607) even it is not crashing animore (or it is just a mater of time) ... should i reopen it?
By: Serge Vecher (serge-v) 2007-02-14 12:31:13.000-0600
KNK, please checkout 1.4 from SVN, there were some issues with SIP transfer that were fixed after ther 1.4.0 release.
By: Kaloyan Kovachev (knk) 2007-02-15 01:35:45.000-0600
Yes it is the latest r54290.
I have discovered this with 1.4.0 (as mentioned in bug 7607), but it is still the same about the timeouts ... the other bug mentioned with Sipuras is gone now.
By: Kaloyan Kovachev (knk) 2007-02-15 06:07:34.000-0600
OK, may rename this to "autofallthru does not default to yes" and the reason was the g flag for Dial without having Hangup() at the end
By: Joshua C. Colp (jcolp) 2007-02-15 09:16:06.000-0600
Since this appears to be a configuration issue I'm closing this out. Peace!
By: Kaloyan Kovachev (knk) 2007-02-15 09:31:07.000-0600
Reopening because it is still a bug:
* The autofallthrough setting introduced in Asterisk 1.2 now defaults to 'yes';
if your dialplan relies on the ability to 'run off the end' of an extension
and wait for a new extension without using WaitExten() to accomplish that,
you will need set autofallthrough to 'no' in your extensions.conf file.
; If autofallthrough is set, then if an extension runs out of
; things to do, it will terminate the call with BUSY, CONGESTION
; or HANGUP depending on Asterisk's best guess. This is the default.
; If autofallthrough is not set, then if an extension runs out of
; things to do, Asterisk will wait for a new extension to be dialed
; (this is the original behavior of Asterisk 1.0 and earlier).
so it is expectecd to have "autofallthrough-yes" by default which is not the case.
Apologies if instead of reopening i should file a new bug. If so just let me know
By: Kaloyan Kovachev (knk) 2007-02-16 06:52:31.000-0600
The patch attached makes aware the pbx for the default values of some global variables when omitted.
autofallthrough is not the only one - writeprotect's default value is also ignored, which may cause much more problems
it wasn't necessary, but just in case i have applied the changes for the 'static' and 'clearglobalvars' defaults ... they may change in future.
By: Joshua C. Colp (jcolp) 2007-02-16 12:15:24.000-0600
Fixed in 1.4 as of revision 54898 and trunk as of revision 54901 by the fabulous Russell. Thanks KNK!