Printer problem on HP-UX 11.0

[Jon Jenkins]

I've recently transfered 40+ printer queues from a 10.20 server to a new 11.0 server.
All but two are working fine. I've deleted the queues for those, re-added the spooler queues, but the result is always the same - prints sit in the queue and go nowhere. I've checked that lpsched is running, check that they are enabled and accepting queues, I've run jetdirect diagnostics and all the information that comes back is that everything is working.

Both printers are HP Laserjet 5N, we have a third identical printer that works.

If I print from the old server all the printers still work.

What can I do next?

[A. Clay Stephenson]

HI:

First thing to check is /var/spool/lp/log for clues.

[Jon Jenkins]

Thanks for the quick response.

The log file only contains lines such as;

Start: printer7-7069 Fri Jul 27 18:15:02 BST 2001

[A. Clay Stephenson]

Hi Jon,

One other thought, since you did not say this explicitly, have you printed a test page from jetadmin? This will allow us to separate printer communication from lp issues.

Clay

[Jon Jenkins]

Hi Clay,

Good point. One quick(ish) run around the building confirms that jetadmin prints sucessfully to both printers (still have prints in queues for both).

Jon

[A. Clay Stephenson]

Okay Jon,

Now we know we are fighting lp. I suspect that you have a couple of hpnpf processes that are hung. First cancel all jobs to these printers.
Next do a ps -ef | grep hpnpf and kill any processes. Do an lpshut and an lpsched. Trying sending a new small text job.

If this fails, cancel the jobs. Then kill the hpnpf's again. I would then delete and re-add the printers from jetadmin. I think that will fix you.

[Victor_5]

Strange! Sounds like it is the lp spooler problem, try the following steps, maybe it can helps.

1. lpshut
2. If /var/spool/lp/FIFO and /var/spool/lp/SCHEDLOCK still exist, remove them
3. Clear the outputq file.
> outputq
4. Kill any lpsched processes that may be reported by:
ps -ef | grep lpsched
5. Restart the scheduler
lpsched
6. Verify that it is running:
ps -ef | grep lpsched
lpstat -t | more

[Jon Jenkins]

Thanks for the replys on Friday (being in the UK, and at the end of the week, I wasn't able to act on the suggestions until today.

Clay, I searched for hpnpf pprocesses, but there weren't any. I'd already deleted and readded the printers a few times in jetadmin, so as I hadn't changed anything I didn't try again.

Shawn, after lpshut I checked for FIFO and SCHEDLOCK, but they weren't there. I typed "> outputq" and restarted the scheduler. Unfortunately, all the print queues then hung, and I went back and deleted and readded the queues in jetadmin. One of the two effected printers also started printing; and the last one while it does print, now prints 20 copies of each print.

Later on this afternoon all printing stopped again, and I discovered that another lpsched was running, killed that one and everything started OK.

So, my only problem now appears to be one printer that prints multiple copies of a single print request. Any more good ideas?
Thanks alot for the help so far.

[Jon Jenkins]

I called HP today to help resolve the this problem. By this morning both printers were printing, but printing the print jobs continually (i.e. the same one page print again and again). We ran through shutting down lpsched (had to kill -9 and delete FIFO and SCHEDLOCK and outputq). Same problem. Rebooted the server (making myself popular), same deal.

So to summarise; 10.20 server prints fine, 11.0 server prints infinite copies (print job remains in spooler queue indefinitely). We have 3 HP Laserjet 5N printers, one working, 2 not. Switched the IP's on the printers, problem stayed with the printer.

The fix? In the PCL Fonts Menu on the working printer the SYM SET was ISO L1, on the two that didn't work SYM SET was PC-8. Once changed to ISO L1 the problem was solved.

This means that 11.0 outputs different control characters from 10.20 that the printers don't understand, which causes completed print jobs not to be deleted from the spooler (I think).

Sorry if this seems rambling but hopefully it may help if anyone else has the same problem.