Summary:ASTERISK-03555: false Hangup detected on bridged Zap calls
Reporter:Matt Florell (mflorell)Labels:
Date Opened:2005-02-21 12:47:55.000-0600Date Closed:2011-06-07 14:10:10
Versions:Frequency of
Environment:Attachments:( 0) messages_Zap-bridge_false_hangup.txt
Description:We have a call redirection system setup inhouse to send calls from an incoming  T1 to an external dialed out number on a PRI:
 Zap(call comes in) -> Asterisk -> Zap(call dials out)

The call comes in on a Robbed-bit T1 (d4/ami EM Wink) and goes out on a PRI. We are using Asterisk release 1.0.5 with a T405P Digium quad T1 card.

Randomly the calls will drop less than a minute into the call. The Debug messages at the end of the call always say something like this: (incoming call on Zap/58, outgoing on Zap/73)

Feb 21 11:37:28 DEBUG[16804]: Exception on 63, channel 58
Feb 21 11:37:28 DEBUG[16804]: Got event On hook(1) on channel 58 (index 0)
Feb 21 11:37:28 DEBUG[16804]: disabled echo cancellation on channel 58
Feb 21 11:37:28 DEBUG[16804]: Unlinking slave 73 from 58

The strange thing is that the person that called in did not hang up, in fact they are usually talking when this call goes dead.  I even have an audio recording of one of these callls as it drops and you can tell it was not a true hangup. Only about 3% of the calls that redirect have this happen to them(34 out of 1000 for a 24 hour period), and it seems to be random as to which ones drop. Calls that dial out on either T1 and calls that come in and are not redirected do not seem to have these problems(at least I haven't heard any complaints).

Going through my messages debug logs it looks like the calls are always answered and always bridge successfully. The conversation time can be anywhere from 3 to 60 seconds before the false on hook is detected and the call is dropped.

This server was previously using an old CVS(2004-08-07) and didn't seem to have this problem before we upgraded it to 1.0.5

I have callprogress=no and busydetect=no in zapata.conf but that doesn't seem to help.

Is there any way to make Asterisk less sensitive to hangups on the robbed-bit T1 line if that's even the cause?

Attached is a call that was hungup after 14 seconds of talk time.
Comments:By: Brian West (bkw918) 2005-02-22 12:54:35.000-0600

We are getting some false hangups too but I can't nail it down to one thing.


By: Mark Spencer (markster) 2005-02-22 23:46:01.000-0600

The only way to get a hangup on a RBS is with the bits actually changing.  Can you watch with zttool to see what is happening?  Also as this is a tech support request, I will probably ask you to move this to support@digium.com instead.

By: Brian West (bkw918) 2005-02-23 00:35:00.000-0600

I think I found our problem... It was related to busy detect.. i'll know in a 24 hours or so.


By: Matt Florell (mflorell) 2005-02-23 09:03:05.000-0600

Is there any way to have the zttool output go to a file across all T1 channels on an Asterisk server? We have 20,000 calls going on this server per day across 2 RBS T1s and one PRI and I really doubt I would catch it if I was looking at it.

Also, I have logged false hangup detects on one of our other servers using RBS T1s from another carrier, So I know it's not the specific machine and it's not the carrier.

This does not seem to happen on our PRIs.

Is this still a support issue or should I leave it here?

bkw - what exactly was your busydetect problem? did you have it enabled?

By: Mark Spencer (markster) 2005-02-26 01:53:12.000-0600

I think support is probably the place to take this.  However if for any reason you don't feel you're getting the answers you need, feel free to contact me directly.