Client unable to print to partner server

[angelb]

Hello all.

I have two CUPS servers(A and B) and 1 Solaris client all running 1.3.8. The
client polls both servers:

# ps -ef|grep -i cups
cups 21146 21144   0 14:12:59 ?   0:01 cups-polld stlam507 631 3600 631
cups 21145 21144   0 14:12:59 ?   0:01 cups-polld stlam506 631 3600 631
root 21144     1   0 14:12:58 ?   2:21 /opt/TWWfsw/cups138/sbin/cupsd -
c /etc/opt/TWWfsw/cups138/cupsd.conf

When I set printer "ricohc" to REJECT on server A, the client's print job is
never printed in server B. Below is a strip from the clients log:

I [12/Mar/2009:14:31:58 +0600] [Job ???] Request file type is text/plain.
I [12/Mar/2009:14:31:58 +0600] [Job 10] File of type text/plain queued by "root".
I [12/Mar/2009:14:31:58 +0600] [Job 10] Started backend /opt/TWWfsw/cups138/lib/backend/ipp (PID 24527)
E [12/Mar/2009:14:31:58 +0600] [Job 10] Print file was not accepted (Destination "ricohc" is not accepting jobs.)!
E [12/Mar/2009:14:31:58 +0600] PID 24527 (/opt/TWWfsw/cups138/lib/backend/ipp) stopped with status 1!

The error is expected and is normal, I think, but I didn't expect the client
to not be able to print through server B.

What I noticed too is that from the client, whe I submit a job, only server A
receives the print jobs and never server B. So, it appears round-robbin feature
isn't working correctly. This may be the same reason why the client is unable
to print to the partner server when the other server is down.

In CUPS 1.2.12, I have this setup in production and works perfectly well where
if one server is down, clients can continue to print through the partner
server.

Has anyone seen the same issue in 1.3.8?

Any suggestion or recommendation on how to fix this is much appreciated.

Thanks.