In our production environnment, I have a printer which users print to from a Unix (AIX) application called iMX (based on Java and Oracle Forms, if that matters) The official documentation page is there http://www.codix.eu/download.php?language=eng if you need any very general info.
The problem is that at some point during the day, generally when the load gets higher, the printer stops printing documents that iMX sends, because one document gets kind of "stuck".
Every of our end-user have :
- A login/password they use to open a session on their Windows XP workstation
- A different login/password they specifically use to connect to iMX once they opened their iMX client
The server side of iMX runs on the AIX server with a generic account 'imxUser', the same which processes print requests from iMX users (see lpstat below).
The faulty printer is also directly reachable from Windows (XP, mostly) clients, ie people can "see" it in their printer manager. They can print Word documents or other things from their workstation, that always work. Please note that this faulty printer is of a different brand than most we use. The supplier has come to check the printer once but nothing came of it.
Lpstat (I removed the Files and User information):
Queue Dev Status Job Files User PP % Blks Cp Rnk ------- ----- --------- --- ------------------ ---------- ---- -- ----- --- --- pName @pName CONNECT 545 xx.user.xx imxUser 25 1 1 : (WARNING) Connection to server timed out after retrying. rembak: errno = 79: Connection refused pName pName HOST_DOWN
- ping pName works okay.
- If 'imxUser' prints something from a terminal with 'lp -d pName filename', it gets queued, and sits forever in the queue, at least until job 545 is manually cancelled (either through 'cancel' in terminal, or by clicking on "reinit printer" on the printer remote web access).
I can't reproduce this problem at will, it is reproducable once or twice per day (when load is high enough). When that happens, the end-users simply configure their workstation to use a different printer as default for iMX, so this problem is not blocking. However, we will eventually have to resort to switch to a printer of a different brand even though we can't even be sure that would solve the problem.
The jobs that get stuck have nothing common between them.
I'm at a loss here, due to my limited knowledge in printers/Unix (I can check things but I can't take the risk of breaking anything on a production server), the very bad reproducibility which relies on end-users' action, and the network environnment and configuration which I'm not too familiar with (not of my doing and I don't have actual control over it).
What would be the right approach here, and the right things to look for ?
VERY important EDIT (sorry I forgot to say that before the first answer) :
- I can unstuck the things if I manually cancel the current job after people told me they can't print anymore, but that has to be done manually and stays unexplained.
- I don't have root access on the server (I only have access to the imxUser, and my own personal user, which can probably do about nothing), if I have to do things as root, I will have to forward them to other people.
PS. Please ask if you need any more information