Job spooling to print server

[Christian Niessner]

Hi all,

I have a little problem setting up a cups setup with distributed servers.

In my setup there are different cups servers. There is one "dedicated" cups server that manages some common, high-troughput printers. The "client servers" also have printers attached and manage some associated printers directly, but also import the high-volume printers from the dedicated server.

On the client servers there get lots of 1-page jobs spooled by a local software. With the local printers everything works fine, but with the imported Printers from the dedicated server there is a problem.

Now I have following issues:

Is there cups 1.1.x running on the client, the troughput is okay. But the jobs get mixed up - they get printed in a different order than they get spooled. And that's a no-go for my use.
I've seen cups spawns lots of ipp-processes each trying to transfer one job to the server. i think that's the cause why the jobs get mixed up.

I searched the forum a bit and it seems that this is a cups 1.1 problem, that should be solved in cups 1.2.

Next try, both machines cups 1.2.12:

The transfer from the 'client' to the 'dedicated' server is very slow - only one job every 10 secs. many to slow, the printer throws 80-100 pages per minute if it gets feeded properly...

I investigated a bit and found that the ipp process on the client 'hangs' for a long time. after having a quick look on the source, i found that it's waiting for the server to complete the job. and in this waiting function, there is a sleep(10)... well...
i discovered another thing: ipp know's an url-option: ?waitjob=no - i think this would solve my problem. Just spool the jobs as fast to the server as possible - and let the server do the work...

But how to tell the server to append this option to all exported printers?

Or can i tell the client to append it for imported printers?

Also the 'compression' option would be nice for remote offices...

Any Hints?

Thank you,
    chris