Summary:ASTERISK-09659: chan_iax2 reports dropping frame from (IP) due to frame already in process during calls
Reporter:xmarksthespot (xmarksthespot)Labels:
Date Opened:2007-06-12 12:19:09Date Closed:2007-06-12 14:38:57
Versions:Frequency of
Environment:Attachments:( 0) iaxdump1.zip
( 1) iaxdump2.zip
( 2) iaxdump3.zip
( 3) iaxverbosedebug1-cleanedstartup.txt
( 4) iaxverbosedebug2-cleanedstartup.txt
( 5) iaxverbosedebug3-cleanedstartup.txt
Description:During iax calls, asterisk produces the messsage:

WARNING[2818]: chan_iax2.c:6322 socket_read: Dropping frame from (callno 32773) of type 6 (subclass 11) due to frame of type 6 (subclass 2) already in process

With varying type and subclass numbers. The connection sounds perfectly fine but this constantly appears in the CLI, albeit seemingly randomly.

This would normally be caused by packet loss, but given the network setup it is doubtful (LAN, between two machines connected directly on the same switch). I've also verified numerous times the network, as well as replaced cabling and changed switch ports, as well as changed switches.

****** STEPS TO REPRODUCE ******

The exact detailed setup is this:

PRI -> Asterisk 1.4.0 -> IAX2 -> Asterisk Branch 1.4 SVN 67804

Physically it's like this:

PRI -> Machine 1 (IAX gateway) -> cable -> switch -> cable -> Machine 2

The machines both have broadcom gigabit ethernet adapters.

To reproduce, I just have to initiate a call to the pri, goes to the SVN machine. Dial an extension, pick up the phone and wait a few seconds, and it starts happening. The message starts appearing over and over again.


Also worth noting is that multiple SIP phones are on the same network and perform corretly, without warnings or error messages. A number of phones are also connected to the machine on which the debug is being taken (Machine 2).

All this is data has been taken on Machine 2.

If you need tcpdumps of both side of the communication, it can be done, but it may take a moment.

Anything else required jsut mention and I will give/do.

I can test all I want on machine 2 (change code, etc), however machine 1 is more delicate as some people depend on it actually.

I am submitting a bunch of files:
iaxverbosedebug1.txt : The usual verbose debug. This is as it happens to me.
iaxdump1 : This is a tcpdump of the communication.
iaxverbosedebug2.txt : Same verbose debug with voicemail disabled for clarity.
iaxdump2 : This is a tcpdump of the second communication.
iaxverbosedebug3.txt : This is verbose with iax2 debug and no voicemail.
iaxdump3 : This is a tcpdump of the third communication.

Thank you very much. Any help is appreciated.
Comments:By: xmarksthespot (xmarksthespot) 2007-06-12 12:28:44

The repoduce information that I submitted does not appear in the report, so I'll say I can reproduce this issue at will, just by making a call.

Otherwise here's how it's set up:

PRI -> Machine 1 (IAX) -> IAX2 line -> Machine 2

Physically it's as such:

PRI -> Machine 1 -> cable -> switch -> cable -> Machine 2

Just needed to reiterate this since it doesn't appear in the bug report for some odd reason.

I submitted verbose debugs with the startup cleaned instead of the full ones.
I also had to zip the tcpdumps as it seemed to crash mantis or something.

By: pj (pj) 2007-06-12 12:34:42

I have same issue, it's starts appearing after latest update, before about a two weeks it working fine.

By: Joshua C. Colp (jcolp) 2007-06-12 14:38:56

Fixed in 1.4 as of revision 69012 & 69014, and trunk as of revision 69013 & 69015.