[Asterisk-Users] PRI deadlock problem is 1.2.1

[Johann]

I thought this problem with PRI and channels getting out of sync was fixed in 
the 1.2.x release of Asterisk.  Here are the errors:

Jan  5 13:59:05 WARNING[1253]: chan_zap.c:8360 pri_dchannel: Ring requested on 
channel 0/2 already in use on span 1.  Hanging up owner.
Jan  5 13:59:11 WARNING[1253]: chan_zap.c:8360 pri_dchannel: Ring requested on 
channel 0/2 already in use on span 1.  Hanging up owner.
Jan  5 13:59:13 WARNING[1253]: chan_zap.c:8360 pri_dchannel: Ring requested on 
channel 0/2 already in use on span 1.  Hanging up owner.
Jan  5 13:59:14 WARNING[1253]: chan_zap.c:8360 pri_dchannel: Ring requested on 
channel 0/5 already in use on span 1.  Hanging up owner.
Jan  5 13:59:15 WARNING[1253]: chan_zap.c:8360 pri_dchannel: Ring requested on 
channel 0/2 already in use on span 1.  Hanging up owner.

Further testing and it shows the same patterns as before.  Using Queues, 
callback agents, and SIP phones.  Once a channel gets out of sync(always one 
with the queue) then no new calls can come in on the PRI line.  Existing calls 
are fine however.  In addition trying to do anything in the CLI dealing with 
queues results in no response and the refusal to do anything else in the CLI and 
it has to be killed and connected again.  Asterisk also can't restart itself 
cleanly in this state and has to be fixed.

We never tracked the problem down and people with similar reports said the 
problem didn't occur in 1.2...and it does us :(

Restarting Asterisk seems about the only way to prevent the deadlock and in a 
production PBX that accepts calls 24/7 that isn't acceptable...


--johann