Summary:	ASTERISK-08199: Asterisk is sending INVAL packages without a reason
Reporter:	Loic Didelot (voipgate)	Labels:
Date Opened:	2006-11-28 05:54:37.000-0600	Date Closed:	2006-12-18 10:07:52.000-0600
Priority:	Major	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:
Description:	According to the IAX protocol a call should be disconnected when INVAL frames are recieved. This is what iaxclient does. Since we upgraded to asterisk 1.4 we noticed that the average call duration went down. So we started investigating. We noticed that at some point (random duration; can happen at any time, normaly less than 10 minutes) asterisk sends 2 INVAL frames to iaxclient and iaxclient hangsup but does not send a HANGUP package. A few seconds after this asterisk stops both legs of the call because it doesn't recieve any voice frames on the iaxclient leg. Now we modified iaxclient to ignore INVAL frames (yep this is ugly) and now it works perfect again. We see again calls with a duration of 2 hours and asteriks still send periodically INVAL frames, but it has no impact on the call. ****** ADDITIONAL INFORMATION ****** We tried different asterisk 1.2 versions and different asterisk 1.4 versions from SVN. The problem does not exist on asterisk 1.2. I guess it has something to do with iax and the multithread implementation.
Comments:	By: pj (pj) 2006-11-28 06:13:13.000-0600 maybe it can also relate to this bug symptoms ... http://bugs.digium.com/view.php?id=8325 http://bugs.digium.com/view.php?id=8273 By: Loic Didelot (voipgate) 2006-11-28 06:13:43.000-0600 There seems to be a relation with ASTERISK-8198 And by the way, we use asterisk realtime with odbc and we do not use iax-trunking. By: Anthony LaMantia (alamantia) 2006-12-04 12:55:35.000-0600 can you upload an IAX2 debug of this issue. By: Anthony LaMantia (alamantia) 2006-12-12 10:14:32.000-0600 voipgate, are you free to post the iax2-debug of the problem? By: Loic Didelot (voipgate) 2006-12-18 06:58:46.000-0600 The problem does no longer occure in revision 48548. By: Serge Vecher (serge-v) 2006-12-18 10:07:51.000-0600 thanks for the note, please feel free to reopen with new IAX debug if the problem resurfaces.