Hung print spooler

[Tony Facca]

(sorry, I deleted the subject line from the original message)

Bill Lasher <psuvm!w0l at psuvax1.cs.psu.edu>  writes:

> We have a room full of Personal IRIS's on a network, with several plotters and
>
>		<< stuff deleted >>
>
> sit there.  If we cancel the job that hung, the queue is freed up and
> everything works fine for awhile.  It doesn't appear to be a specific file; if
> a file gets hung and we cancel it, then re-submit the same job, it works fine.
> We have contacted the SGI hotline, but they don't yet know what the problem is
> however, they did say that someone else had reported it.  Anyone else having
> this problem; or any ideas as to what might be causing it?
> 
> Many thanks,
> 
> Bill Lasher
> 

I reported the same problem way back when 3.1 first came out.  The problem
didn't exist in 3.0 but started when I upgraded to 3.1   I called the hotline
and the guys worked on it for several days but couldn't find exactly what the
problem was.  Seems that there is a "sleep" in the network interface script
which causes the spooler to appear hung.  This section of code is executed
when an attempt to spool a file to a machine which is down (or not accepting
requests) is made.  The current job is held up, and the remote copy will be
tried again later.  I suppose this is preferable to disabling the local print
spooler as was done in the past.  If you look at the script you'll see that
the lines containing the "disable" command are commented out as well as the
"exit 1" command -- so there is no way out of this section of code if the
file is not transferred.   I don't know why it doesn't work eventually (when
the target machine comes back up)?

Anyway, I just used the 3.0 script in place of the 3.1 stuff.  This works fine.
And, since I had several other changes in the 3.0 script, it was easier than
trying to get the 3.1 version working.  I guess this is where the Hotline call
must have died.  Maybe I canceled it, I don't remember.   

The original script used by "mknetpr" is found in /usr/spool/lp/etc/lib and is
called "netface".  This is NOT the one I change.  Instead, I let mknetpr do
its thing, then just swap out the final script in /usr/spool/lp/interface for
the old script.  Rather than post the script here, you can contact me directly
and I can let you "ftp" it from me.  

Good luck.

--
-----------------------------------------------------------------------------
Tony Facca                     |     phone: 216-433-8318
NASA Lewis Research Center     |    
Cleveland, Ohio  44135         |     email: fsfacca at lerc08.lerc.nasa.gov
-----------------------------------------------------------------------------