Summary:ASTERISK-05080: Calls to Voicemail via IAX drop (ast_play_and_record: No audio available)
Reporter:Trevor Hammonds (trevmeister)Labels:
Date Opened:2005-09-14 05:54:19Date Closed:2011-06-07 14:10:06
Versions:Frequency of
Description:Calls are dropped by app_voicemail without taking a message under the following scenario:

Call originates on System A (lsanca01) via PRI.  System A sends call to System B (snanca01) via IAX, which sends call to System C (msvjca01) via IAX.  System C rings a SIP device until timeout, then System C's voicemail application kicks in.  This causes System B to drop out of the call shortly after the voice mail message begins playing.  Shortly after this happens, the entire call is dropped -- always immediately after playing the beep.  This results in an empty voice mail message, and no immediate indication to the caller that they did not actually have their message recorded.  

Also, no audio is sent from the originating leg of the call while the OGM is playing (e.g. if the caller presses #, *, 0, etc., nothing breaks the outgoing audio stream).

The problem will not happen if there is no Dial command executed on System C prior to the Voicemail command (e.g. if I NoOp the Dial command, or replace it with 20 seconds of ringing, everything works as expected)

To solve this problem, I have re-routed calls from System A directly to System B.  This, however, is not a desirable solution.  

I will greatly appreciate any assistance.  If there is anything you need me to do, please let me know.  


I am running CVS HEAD as of 2005-09-14, and have noticed this for about a week.  

A previous message on the dev list in June indicates a nearly identical situation.  Since his message was fairly detailed, you may wish to refer to that while investigating.  

See http://lists.digium.com/pipermail/asterisk-dev/2005-June/013510.html

Typical error message on Server C:
WARNING[9403]: app.c:631 ast_play_and_record: No audio available on IAX2/snanca01-16384??

Typical messages displayed on Server B's console:
   -- Releasing IAX2/msvjca01-5 and IAX2/lsanca01-4
   -- Hungup 'IAX2/msvjca01-5'
 == Spawn extension (macro-transfer-msvjca01, s, 3) exited non-zero on 'IAX2/lsanca-4' in macro 'transfer-msvjca01'
 == Spawn extension (inbound, 1xxxxxxxxxx, 1) exited non-zero on 'IAX2/lsanca-4'
   -- Hungup 'IAX2/lsanca01-4'
Comments:By: Michael Jerris (mikej) 2005-09-14 07:59:47

Not major as there is a workaround.  also, try notransfer=yes.  This may be related to ASTERISK-4798

By: Trevor Hammonds (trevmeister) 2005-09-14 14:56:49

To what workaround are you referring?  I did not see anything in 4924 other than using 'notransfer=yes', which is a bad idea in our situation, as successive transfers will eat up increasing bandwidth.

Also, there is NO workaround for the fact that no caller audio is being passed to the final destination.

By: Michael Jerris (mikej) 2005-09-14 15:01:12

The workaround is directing the call directly from A to C.  Testing with notransfer=yes was a request to test in that configuration to verify it corrects the issue.

By: Michael Jerris (mikej) 2005-09-14 15:01:56

Also, we need full debugs of the call from all 3 servers.

By: Trevor Hammonds (trevmeister) 2005-09-14 15:19:35

I do not have access to the server with the PRIs.  I will post debugs from the other two, however.  Do you just need iax2 debug output?

By: Kevin P. Fleming (kpfleming) 2005-09-14 19:08:59

'set verbose 10', 'set debug 10' and 'iax2 debug', and capture the results in the 'full' log file please. Thanks!

By: Clod Patry (junky) 2005-10-04 22:06:57

Trevmeister: we still waiting your debug here.

By: Michael Jerris (mikej) 2005-10-04 23:10:22

suspended due to no response.  Unfortunately we can not debug this issue further without additional information.  Please re-open this bug when you have the appropriate information available.