asterisk users - Oct 2008 - Sip Trunking

I have several branch offices, each with their own Asterisk server 
(version 1.4.22.1) handling their PBX functions.  All of these offices 
need to talk to each other.  In sip.conf I created a peer entry for each 
office with a username of branch-user and a friend entry for every 
branch-user with the username being just the branch, for example:

[Office2]
username=Office1-user
host=10.10.80.253
type=peer
context=internal
secret=sipptrunk
dtmfmode=rfc2833
disallow=all
allow=ulaw
call-limit=20

[Office1-user]
username=Office2
host=10.10.60.253
type=user
context=internal
secret=sipptrunk
dtmfmode=rfc2833
disallow=all
allow=ulaw
call-limit=20

This mostly works.  I am able to place calls between all of the branches 
about 90% of the time.  What is happening is that if no call has been 
placed between two of the branches for a while then the next call that 
is made may get a "busy/congested" message like follows:

-- Executing [6042 at internal:1] Dial("SIP/21-00a16c70", 
"SIP/42 at Office1||Tt") in new stack
    -- Called 42 at Office1
    -- SIP/Office1-009f49b0 is circuit-busy
  == Everyone is busy/congested at this time (1:0/1/0)
    -- Executing [6042 at internal:2] Hangup("SIP/21-00a16c70",
"") in new
stack

If I try a few more times, the call will eventually go through.

I have tried setting qualify=yes in sip.conf, and also setting up a 
constant ping to make sure there were no connectivity issues between the 
two sites.  Any idea what could be causing the problem?

Thanks,
Brent Davidson

Brent Davidson <brent at texascountrytitle.com> writes:
> I have several branch offices, each with their own Asterisk server 
> (version 1.4.22.1) handling their PBX functions.  All of these offices 
> need to talk to each other.  In sip.conf I created a peer entry for each 
> office with a username of branch-user and a friend entry for every 
> branch-user with the username being just the branch, for example:
You should only need peer entries... type=user is dying.


/Benny