[Asterisk-Users] Resolving timing issues with dual PRIs in a TE411P

[Anthony Rodgers]

Greetings, everyone, and Happy New Year!

I have a question relating to running two PRIs into a single TE411P. We 
have been experiencing echo, noisy MOH, poor audio call quality and so 
forth that started at around the time we introduced the second PRI into 
the equation. Here is our zaptel.conf:

span=1,2,0,esf,b8zs
bchan=1-23
dchan=24
#clear=1-24
loadzone = us
defaultzone=us

span=2,1,0,esf,b8zs
bchan=25-27
dchan=48
#clear=1-24
loadzone = us
defaultzone=us

And here is our zapata.conf:

[trunkgroups]

[channels]

context=incoming
usecallingpres=yes
echocancel=128
callerid=asreceived
usecallerid=yes
echocancelwhenbridged=yes
echotraining=500

; Nortel PBX
rxgain=2.0
txgain=-17.0
group=1
callgroup=1
pickupgroup=1
switchtype=national
signalling=pri_net
facilityenable=yes
channel => 1-23

; Allstream PRI
rxgain=-2.0
txgain=-17.0
group=2
callgroup=2
pickupgroup=2
switchtype=dms100
signalling=pri_cpe
facilityenable=no
channel => 25-27

musiconhold=default

We are master to the Nortel PBX, which gets its clocking from Telus (a 
telco), and slave to the Allstream (another telco), from where the 
Asterisk gets its clocking.

We have been through the IRQ/APIC/hyperthreading thing already, and our 
gain levels were set using ztmonitor. zttest gives Best: 100.000000 -- 
Worst: 99.987793 -- Average: 99.998923 after 885 passes. We have 
another, identical server that plays MOH with no problems without a PRI 
attached (presumably using internal clocking), and that same server and 
another similar server had no issues for months with only one PRI 
attached (one had the Nortel and one had the Allstream, with an IAX 
trunk between them).

Is it possible that the two PRIs have timing differences between them 
that could cause problems? If so, how best could we resolve them?

Regards,
-- 
Anthony Rodgers
Business Systems Analyst
District of North Vancouver
Web: http://www.dnv.org
RSS Feed: http://www.dnv.org/rss.asp