asterisk users - Feb 2010 - Intermittent DAHDI issue with a PRI line causing asterisk to crash!

Hello all,

I've got an intermittent issue with my asterisk set-up, and I'm pulling 
my hair out!

Mostly everything works fine, but I get an error once every few days 
that sometimes (but not always) causes asterisk to stop accepting new 
calls and also stop responding to commands on the console (asterisk -r).



The usual errors look something like this:

[Feb 19 13:09:52] ERROR[18728] chan_zap.c: XXX Missing handling for mandatory IE
4 (cs0, Bearer Capability) XXX
[Feb 19 13:09:52] ERROR[18728] chan_zap.c: XXX Missing handling for mandatory IE
24 (cs0, Channel Identification) XXX
[Feb 19 13:09:53] WARNING[18728] chan_zap.c: Duplicate setup requested on
channel 0/2 already in use on span 1

We were running an old version of asterisk (1.4), so I upgraded to 1.6 to see if
the issues would resolve by running the latest stable version.

Due to the upgrade, the error message changed from zap to dahdi!

[Feb 23 11:45:39] ERROR[15861] chan_dahdi.c: XXX Missing handling for mandatory
IE 4 (cs0, Bearer Capability) XXX
[Feb 23 11:45:39] ERROR[15861] chan_dahdi.c: XXX Missing handling for mandatory
IE 24 (cs0, Channel Identification) XXX
[Feb 23 11:45:58] ERROR[2418] chan_dahdi.c: Can't destroy call 1807007664!

After these messages, asterisk would only accept up to 8 calls (we have a single
e1 for up to 30 channels).  When the 9th caller called, they would get an
engaged tone.

Running 
> core show channels verbose
while asterisk in this state causes the asterisk console client (asterisk -r) to
ignore me after printing out /most/ of the expected output.  It crashes just
before the totals IIRC.  When this has happened in the past, I have had to
killall -9 asterisk.  This time I was able to issue a 'core restart
gracefully', but one of the channels seemed to be stuck in the up position. 
I'm afraid I haven't got more debugging info as this is a live service
and needed restarting ASAP.

Today I had a similar crash:

    -- Hungup 'DAHDI/0:5-1'
[Feb 25 10:42:09] NOTICE[11222]: pbx_spool.c:349 attempt_thread: Call completed
to DAHDI/g1/07973734513
[Feb 25 10:42:09] WARNING[7827]: chan_dahdi.c:12124 pri_fixup_principle: Call
specified, but not found?
[Feb 25 10:42:09] WARNING[7827]: chan_dahdi.c:13293 pri_dchannel: Hangup on bad
channel 0/5 on span 1


One thing that is interesting here is that it's refering to channel 
'DAHDI/0:5-1', when the channels are normally referred to as 
'DAHDI/5-1', 'DAHDI/3-1' etc.  Asterisk continued to work
normally after
this error and did not require a restart.

I haven't found a way to reproduce the fault yet.  The fault seems to 
happen proportionally more often as call volumes rise and fall.



So does anyone have any clues as to what's going on here? 
What would the appropriate debugging for this be?  As it happens so 
rarely there's a tradeoff between filling the logs with jam and figuring 
out what's going on.
Is this likely to be an asterisk issue, or lower level?

Configuration details pasted below.

Thank you for your time.

Kind regards,

James



Asterisk 1.6.2.4 built by root @ roict.co.uk on a x86_64 running Linux 
on 2010-02-21 23:20:17 UTC
dahdi-linux-complete-2.2.1+2.2.1
libpri-1.4.10.2
wanpipe-3.5.10
linux 2.6.18

Supplier: Thus (UK)

[root at sms-1 asterisk]# cat chan_dahdi.conf
;autogenerated by /usr/sbin/wancfg_dahdi do not hand edit
;autogenrated on 2010-02-21
;Dahdi Channels Configurations
;For detailed Dahdi options, view /etc/asterisk/chan_dahdi.conf.bak

[trunkgroups]

[channels]
context=default
usecallerid=yes
hidecallerid=no
callwaiting=yes
usecallingpres=yes
callwaitingcallerid=yes
threewaycalling=yes
transfer=yes
canpark=yes
cancallforward=yes
callreturn=yes
echocancel=yes
echocancelwhenbridged=yes
relaxdtmf=yes
rxgain=0.0
txgain=0.0
group=1
callgroup=1
pickupgroup=1
immediate=no

;Sangoma A101 port 1 [slot:2 bus:3 span:1] <wanpipe1>
switchtype=euroisdn
context=incoming
group=1
echocancel=yes
signalling=pri_cpe
channel =>1-15,17-31