Summary:	ASTERISK-01312: FXO_KS signalled Zap Channels on Adtran 750 Channel Bank Stuck in Rsrvd State
Reporter:	sbisker (sbisker)	Labels:
Date Opened:	2004-03-30 13:25:59.000-0600	Date Closed:	2004-09-25 02:52:14
Priority:	Major	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:
Description:	Periodcally through the day, several Zap channels are getting stuck in a Rsrvd State.  Looks like the hangup isn't being registered properly.  I'm running Zaptel 0.9.0 and asterisk release candidate CVS 3/28/04.  I just upgraded from Zaptel CVS 3/9/04 and Asterisk 0.7.1 and this problem showed up yesterday.  Soft hangup doesn't work, only lifting the receiver and hitting the hook fixes the channel.  The only thing that is different is that I'm no longer using -DOLD_DSP_ROUTINES. ****** ADDITIONAL INFORMATION ****** I've also noted a bunch of these warnings in the console. Mar 30 13:11:44 WARNING[163851]: chan_zap.c:1193 zt_set_hook:  zt hook failed: Device or resource busy
Comments:	By: sbisker (sbisker) 2004-03-31 15:01:11.000-0600 Switched back to 0.7.1 with -DOLD_DSP_ROUTINES and the problem went away.  Will try with CVS and -DOLD_DSP_ROUTINES, but I guess this should be moved to "Core Asterisk"  since it doesn't seem to be a Zaptel problem By: screel (screel) 2004-04-01 15:19:07.000-0600 For what it's worth - I see this all the time as well.  (Also with fxoks signalled channels on two Adtran 750s).  I've also seen that the only way to resolve it is to take the line off hook and hang it back up.  I can provide additional information if it would be of any help... By: Mark Spencer (markster) 2004-04-05 10:00:52 Do you have a way to duplicate this problem at will?  I have a sneaking suspicion it has to do somehow with it acidently making a three way call just when going back on hook.  Also, if you don't need three way calling, try turning it off and see if that makes the problem go away. By: sbisker (sbisker) 2004-04-05 10:03:31 Three-way calling is disabled on all my zap channels as is call-waiting.  I have 3 channels in Rsrvd state right now, and it happens all day. By: Mark Spencer (markster) 2004-04-05 10:04:59 Find me on IRC in about 20 minutes when I get into the office, I'll need login info and then I'll take a look and see if I can figure it out. By: sbisker (sbisker) 2004-04-05 10:05:16 Here's a zap show channel  for a locked channel. File Descriptor: 141 Span: 4 Extension: Context: local Caller ID string: "Ace Ventura" <5290> Destroy: 0 Signalling Type: FXO Kewlstart Owner: Zap/78-1 Real: Zap/78-1 Callwait: <None> Threeway: <None> Confno: -1 Propagated Conference: -1 Real in conference: 0 DSP: yes Relax DTMF: no Dialing/CallwaitCAS: 0/0 Default law: ulaw Fax Handled: no Pulse phone: no Echo Cancellation: 128 taps, currently OFF Actual Confinfo: Num/0, Mode/0x0000 Actual Confmute: No By: sbisker (sbisker) 2004-04-05 10:21:09 I'll find you on IRC. -sb By: sbisker (sbisker) 2004-04-05 16:43:48 Mark.  Let me know if you need to get back on that box tomorrow, or if you've come up with any potential issues? -sb By: twisted (twisted) 2004-04-06 03:12:38 I've also noticed this, but I'm not on an adtran channelbank.. I'm on a Zhone Zplex-10B 24-FXS.  Using fxo_ks signaling, every once in awhile, i will get  a channel stuck in the reserved state, until i physically pick up and hangup the reciever.. This makes me think it's not really the channelbank, but possibly the driver modules..? I'm totally lost on this one, but I can confirm this issue.   I'm using CVS -head as of 4/05/04 for *, and zaptel as of 04/03/04. By: sbisker (sbisker) 2004-04-06 08:13:11 It doesn't appear to be driver related from my viewpoint.  As asterisk 0.7.1 doesn't exhibit the behavior, but asterisk v1.0 CVS 3/28/04 does exhibit the behavior.  Is it only fxo_ks signalling that has the problem? How about fxo_ls or fxo_gs?  I did a source diff from my 0.7.1 and the v1.0 CVS and all in all the only potentially related areas with significant change were channel.c dsp.c and pbx.c By: jharragi (jharragi) 2004-04-06 10:10:35 I also experience this on one of my systems (*,T100P<=>zhone<=>merlin 1030). This may be a clue: The problem seems to be more pronounced on the zhone channels that are plugged into the merlin's line ports than channels plugged into 2500 sets. The merlin takes a noticeably large fraction of a second to respond to changed states so may not look like a typical 2500 set (to the channel bank and *). It may be related to timing issues of the device - or the way it uses the fxs port. Something which can occur frequently in the asterisk <=> key_system setup is pulling dial tone for a short duration. A user may pick up the handset (on the default line) then moments later press a key for another line (hanging up on the default). By: Mark Spencer (markster) 2004-04-06 11:03:42 Try to get me on IRC again today.  I'll probably have to add some more debugging. By: robf (robf) 2004-04-06 11:25:18 I was following another similar bug related to stuck Zap channels, and I thought this problem was fixed in the latest CVS (default branch).  I have a T100P and an Adtran 750.  I could pick up an extension, dial VoiceMailMain ("enter mailbox number"), then hang up, and it would leave the channel "Reserved".  I'm not experiencing that any more, however...  I just upgraded again to the latest CVS just minutes ago.  I'll post here if I start having the problem again... By: Mark Spencer (markster) 2004-04-06 15:02:08 Can you get me an updated zap show channel on the RESERVED channel?  I've added a new feature. By: sbisker (sbisker) 2004-04-06 15:17:50 I'll update CVS tonite and do zap show channel for a stuck channel tomorrow. By: sbisker (sbisker) 2004-04-07 09:12:31 Updated release CVS last night.  Here's a zap show channel.  I don't see anything new in the display. [root@isdpbx01 apps]# asterisk -rx "zap show channel 73" Channel: 73 File Descriptor: 138 Span: 4 Extension: Context: longdistance Caller ID string: "XXXXXX XXXXXX" <5279> Destroy: 0 Signalling Type: FXO Kewlstart Owner: Zap/73-1 Real: Zap/73-1 Callwait: <None> Threeway: <None> Confno: -1 Propagated Conference: -1 Real in conference: 0 DSP: yes Relax DTMF: no Dialing/CallwaitCAS: 0/0 Default law: ulaw Fax Handled: no Pulse phone: no Echo Cancellation: 128 taps, currently OFF Actual Confinfo: Num/0, Mode/0x0000 Actual Confmute: No    -- Remote UNIX connection By: sbisker (sbisker) 2004-04-07 09:15:51 Here's another interesting tidbit of info.  These Zap channels that are getting stuck are getting stuck without the end-user doing anything.  It appears that all of the phones that are getting stuck have their time set via the phone system.  The go on-hook to set time and date, check for MWI, etc.  Then go off-hook.  It looks like during this process, the channels are getting stuck.  Maybe the timing for these operations is quick enough that the hangup isn't registering properly? By: screel (screel) 2004-04-07 09:47:33 The times I'm seeing this are when the line is taken off hook and then back on hook rather quickly.  Just picking up the phone and quickly putting it back on hook seems to repeatedly cause it.  The console shows the channel off hook, then on, then back off again, where it sits indefinitely. By: Mark Spencer (markster) 2004-04-07 10:06:36 That's not an output from latest CVS.  Latest CVS should print something like this at the bottom: Actual Hookstate: Onhook Also be sure you're using CVS head not CVS stable! By: sbisker (sbisker) 2004-04-07 10:24:39 How much of a difference is ther between CVS stable and head at this point?  I'm a little hesitant to blow away my v1.0 stable production system with head CVS as this is the only major issue I'm having.  I don't want to introduce anything new into the mix.  Do you have a patch that I could apply to stable? By: Mark Spencer (markster) 2004-04-07 11:11:17 Just find me on IRC so I can login and make the change. By: Mark Spencer (markster) 2004-04-07 11:45:01 n/m i was able to duplicate it here.  It's definitely a bug in zaptel, so it won't matter if you're using head or stable.  I'll work on it and post some more info when the fix is ready. By: Mark Spencer (markster) 2004-04-08 04:03:24 Fixed in <b>zaptel</b> CVS.  This is actually zaptel bug and not an asterisk bug.