[Asterisk-Users] FXO module in TDM400P (UK, BT) - Hangup detection failing

[StrUK]

Hi,

I'm a UK-based * newbie with a BT line on a PIII 2GHz 1Gb Fedora Core 2  
linux box, fully patched up, experiencing a little difficulty with *  
(both CVS V1-0 and yesterday's CVS HEAD). I've been through the  
archives and even popped on to freenode/#asterisk for a while, but no  
firm resolution presented itself. Last relevant post appears to have an  
unresolved cliffhanger:  
http://lists.digium.com/pipermail/asterisk-users/2004-September/ 
061518.html

I had an X100P as my PSTN link and that was working fine for taking and  
making calls, but of course no caller ID (Plenty of "Didn't finish  
Caller-ID spill.  Cancelling" and that's about it).

Needing CID for my application, I procured an FXO module for my TDM400P  
which already had one FXS module in it (also working fine, happily  
shoving and receiving CID info around between my SIP soft phones and  
the Zyxel SIP hardphone).

The wcfxs module is correctly installed with opermode=uk (dmesg  
confirms this), and polarity CID detection works a treat. Call routing  
is fine and, naturally, none of the SIP stuff has broken :-)

HOWEVER, the FXO module isn't detecting remote party clear down events.

Two example cases:

Zap/4-1 (the FXO module on the TDM400P) detects ring because someone is  
dialling my BT line. My dialplan displays CID info to console for  
debugging and then dials my SIP softphone (SIP/stripes), SIP hardphone  
(SIP/zyxel) and Zap hardphone (Zap/1-1).

1/ If the incoming party clears down before one of the internal  
extensions answers, they continue to ring (* hasn't logged the remote  
hang up event on Zap/4-1) and dead air is heard should they  
subsequently answer.

2/ If the incoming party stays on long enough for voicemail to kick in,  
then I get a recording comprising their message, pop/click, 2secs  
silence, 5secs solid tone, pop/click, followed by dead air (and silence  
detection is the only thing stopping the message hitting the 120 second  
maximum message length ceiling)

Both of these cases are consistent with how I understand * to work  
should the channel not notice Zap/4-1 going down - bully for me, bummer  
for operational use ;-)


Speaking with the ever-so helpful folk on freenode/#asterisk, I've  
tried loopstart signalling (nice idea - failed miserably where remote  
clear down made * think that the remote party was trying to transfer  
calls - lots of MOH messages and 's' priority of the context invoked);  
I've tried tweaking indications.conf to make *'s notion of 'busy' for  
my line match what is being heard as the tone, but that hasn't worked  
and incoming calls still stay 'connected' until the voicemail's silence  
detection cuts them off.

I can live with voice messages being 9 seconds longer than they need to  
be; I can't, though, live with incoming callers that hangup before  
anything answers still resulting in the internal extensions being  
signalled.

I don't have a lighted keypad, but have seen a deflection on a  
multimeter across A/B wire to suggest that the phone company are  
signalling remote clear down with a polarity reversal.  I'm told  
(again, kind folk on irc) that BT do use supervised disconnects and so  
I am doing the right thing using fxs_ks signalling.

I guess my question is: does anyone have polarity reversal hangup  
detection working on a BT line with an fxo module in a TDM400P?

If so - your direction would be most greatly appreciated.   I've posted  
slightly censored versions of my config to http://ermy.net/senast/, in  
case that might help.


Many thanks in advance,

Mark/

--
// if it doesn't go woof when you light it, you ain't dun it right.