Cups problem. "ReadClient() 5 IPP Read Error!"

[Steve Bergman]

I have an ongoing problem with one of my clients' printers.  It's an HP
laserjet 5.  It is connected to an FC1 workstation parallel.  It prints
mostly ascii reports and an occasional email or pdf from gnome.  All
jobs are queued to it from the FC2 server that is running the accounting
programs and everyone's desktop.  It mostly works OK.  Except that every
day or two cups will spontaneously disable it.  This seems to occur
after a particular job prints successfully.  i.e. when I reenable the
printer, the job prints again, in its entirety.  I have set up a cron
job on the FC2 server to watch for that disabled condition every 60 sec
and reenable the printer.  For the most part, they get an occasional
duplicate print job.  However, occasionally, the printer will hang, and
reenabling does not help.  At these times there is a cups "parallel"
backend process trying to send to the printer, forever, but not doing
anything. The printer shows to be "ready".

I've replaced the FC1 box twice, the printer once, and the printer cable
twice.

>From the cups error_log on the FC1 box:

I [18/Aug/2004:13:01:25 -0500] Adding start banner page "none" to job
859.
I [18/Aug/2004:13:01:25 -0500] Adding end banner page "none" to job 859.
I [18/Aug/2004:13:01:25 -0500] Job 859 queued on 'okc_svclaser' by
'holly'.
I [18/Aug/2004:13:01:25 -0500] Started
backend /usr/lib/cups/backend/parallel (PID 1664) for job 859.
E [18/Aug/2004:13:01:25 -0500] ReadClient() 5 IPP Read Error!

I see this when the printer gets disabled.  So what is a "5 IPP read
error?"

On the FC2 server, I have a similar situation, where a job will just
hang going out the parallel port to an HP 4050 right in the middle of a
job.

I'm wondering if parport in Fedora is not ready for real use?

One other question.  When cups disables a printer, why does it disable
it permanently.  Why no retry after 5 minutes or something?

-Steve Bergman