Topic: Mac print jobs failing

[Robert Devantier]

SME Server v5 Update 2, new install.  PII-333MHz, 256MB memory, 4GB SCSI drive, 100Mbit Intel Server adapter.  No users using it yet, plenty of CPU/network/disk disk available.

I have created a few printers, and have Mac users trying to print to them.  The easy jobs come out OK, no problem.  But I have these large jobs going to an HP DesignJet 800 that keep failing.  The plotter will say that it is flushing the rest of the job, then switch to saying it is receivng an HP/GL file.

I have tried Charlie's trick recommended in the developer forum of putting a "-b -h" in the papd.conf file.  This didn't work.  I also tried removing the pipe on the lpr command.  I also brought the PPD over from the Mac, ran dos2unix on it, and specified it in the papd.conf file as well.  On the Mac side, I tried changing the picture encoding from binary to ASCII.  None of these seem to make any difference.

If the users print straight to the plotter's Appletalk name, it comes through just fine.  But with 40-some printers, I'd like to have them on queues.

I will also point out that I run into these same problems with jobs going through my Windows NT 4.0 Server queues.  But I don't expect help on those...

Can anybody offer some assistance?  I've been pouring over the usenet archives, sourceforge (netatalk section), and the e-smith area.  My eyes are tired from reading so much.

Robert Devantier

[Chris O'Donovan]

I have been having very similar problems with SME v5u2. We have six macs and an HP printer, and it has been very difficult to find any commonality between which jobs cause problems and which don't. Changing Mac-Illustrator to ASCII from binary in the Printer-Setup on the mac seemed to help a bit.

Since there are only two printers we have decided to go with $US99 ethernet print servers since I had already spent more than $200 of my time trying to get this working. With 40 printers I would still be looking for an answer.

Looking around just now I found references to the "psf" print filter, which is part of the papd package. The man page for this filter contains several references to HP printers such as "This  kludge  is  a  work-around  for  printers,  notably Hewlett-Packard's LaserJet IV, which will...". I would suggest trying this filter in the papd.conf file.

My macs are at a remote site that I only visit for a couple of hours each week and I have no macs locally so it is difficult for me to try this suggestion myself until the middle of next week. I would be interested in hearing any reports of success, since we haven't actually bought the print servers yet.

Chris

[Robert Devantier]

Chris -

Thanks for the help.  I'm not sure this is what I am looking for.  Looking at the man pages for psf (who decided that v5 shouldn't have the man rpm???) doesn't give me high hopes.  The first kludge is for accounting, which I don't believe is on.  The second kludge is also for accounting.  I'll look into these a little more, but I'm not sure they are going to help.

I updated the LPRng system from 3.7.4 to 3.8.0 this morning.  Once I double-check all the papd.conf and printcap entries to make sure they are how they SHOULD be, I'll send a test print and reply back here.

[Robert Devantier]

Well, things have improved but not gotten completely better.

Updating to the LPRng 3.8.0-1 release was good to do, binary postscript works better but not 100%.

There is a newer netatalk than 1.5pre6-1rh7, and that is pre8.  One of the fixes for the pre8 is for binary postscript.  Unfortunately, they only have the source tarball, no redhat rpm yet.

I still have a couple of print samples that will not print via a spooler (NT or Unix) but print directly to the AppleTalk printer.  My next step is to try printing to E-smith via the LPR option in LaserWriter 8.5.1 and newer.  If that can print the job (and maybe even do it in binary) I'll stop worrying about all of this.

And for anybody who cares, Unix ROCKS for large print jobs.  I had a 36x90 poster going to a DesignJet 800PS.  From clicking "Print" to start of printing, it was 40 minutes on E-Smith (PII-333MHz, 256MB ram) and 120 minutes on an NT4.0 print server (Dual PIII-450MHz, 512MB ram).  Both servers used Intel Pro/100+ Server NICS.  Obvious speed-up comes from NT4.0 having to run the job through its own printer driver.