Recovery of Hung Printer Queues
							9/27/95, JMB
	The normal Laserjet printer queues on the UNDHEP VAXcluster occasionally
hang between jobs.  You will notice this by a job that is "printing" for a
considerable time but nothing has come out, and the printer is OK from the 
IBM/SGI side.  You will also find that jobs queued to the Unix names of the
printers will work, but not to the VAX names.  This is due to a problem in
the DEC Pathworks for Macintosh software that runs the VAX queues.  Because
of this, a batch job RESTARTLJ has been created and runs on UNDHD0 every 
6 hours to restart the queues.  If the printer hangs part way thru a job or
on a job from Unix, this is NOT the problem.

	If you suspect this problem and Jim is not around, the job may be
released by hand.  It will run and requeue itself 6 hours later.  If this
does not clear the problem, running this job does not hurt anything or lose
print jobs.  To release the job, log into the HEPoperator account and do:
$ SHOW QUEUE/ALL/BY=TIME
$ SET ENTRY/REL ###     ! ### is the job entry number from the Show Queue

	VAX queues to Unix printers are fed thru the UCX internet package.
Do SHO QUEUE UCX* to see if the queue is up.  If so, the queue may be stopped
on the Unix end; see below.  If the UCX LPR queue is stopped, it may be
restarted with the START/QUEUE queuename command, or with the command
@SYS$MANAGER:UCX$LPD_STARTUP, again from the HEPoperator account.

	If a print job from Unix fails to start, it may be due to some problem
that stops the print queue either on the executing node or the queuing node.
On the IBMs use the command lpstat to check if the queue is READY or DOWN.
Queues listed with 2 entries are server queues for devices on another node.
You should check the queue on the other node to see which system needs the
queue restarted.  Both the device queues on the executing node and the remote
server queues can stop independently and may need restarted.  A queue may be
working on one node and not on another; this indicates the server queue is down.
If the queue is down on all nodes, the executing queue is down or the device
itself is down.  If the device is down, Unix does not support requeuing jobs
to another printer, as VMS does.

	To restart a queue on IBM, use the root account and start the Smit 
utility.  Select Spooler, then Manage Local Printer Subsystem, then Local 
Printer Queues, then Start a Queue.  Then use F4 or -4 to list the 
possible queues to start.  Use the arrow keys to select the printer to start, 
and hit return to select the printer, and return again to do it.  Exit Smit 
with F10.  If you are running the X version of Smit, use the mouse for all 
selections.  Then check the queue again with lpstat.  If it does not restart, 
check the printer itself for a problem.  To restart a queue on the SGIs, 
check the administration manuals or find Dave or Jim.