How to print from Windows XP to cups-lpd

[Kai Lanz]

Short version: How do I print from a Windows XP client to cups-lpd?

We have RHEL 4 with CUPS v1.1.22. CUPS and cups-lpd are configured and working;
I can print from a Mac OSX client using lpd-style printing just fine. But I
can't print from a Windows XP Professional client, even though it's set up to
print to the same host, same port (515), using LPR, to the same queue.

Mysteriously, the job reaches the server and then seems to evaporate. The logs
show the client contacting cups-lpd via xinetd; here's /var/log/secure:

Sep 4 14:07:18 pangea2 xinetd[5663]: START: printer pid=2219 from=171.64.175.30

Then in /var/log/messages, we see an error from cups-lpd:

Sep  4 14:07:18 pangea2 cups-lpd[2219]: Connection from a65pc20 (171.64.175.30)
Sep  4 14:07:18 pangea2 cups-lpd[2219]: Receive print job for a65-laserjet
Sep  4 14:07:18 pangea2 cups-lpd[2219]: Error while reading file - Success
Sep  4 14:07:18 pangea2 cups-lpd[2219]: Closing connection

The cups/access_log file shows a small flurry of connections:

localhost - - [04/Sep/2009:14:07:18 -0700] "POST / HTTP/1.1" 200 137
localhost - - [04/Sep/2009:14:07:18 -0700] "POST / HTTP/1.1" 200 137
localhost - - [04/Sep/2009:14:07:18 -0700] "POST / HTTP/1.1" 200 77
localhost - - [04/Sep/2009:14:07:18 -0700] "POST / HTTP/1.1" 200 180

And here's the error_log, with logging level at debug:

D [04/Sep/2009:14:07:18 -0700] AcceptClient: 6 from localhost:631.
D [04/Sep/2009:14:07:18 -0700] ReadClient: 6 POST / HTTP/1.1
D [04/Sep/2009:14:07:18 -0700] ProcessIPPRequest: 6 status_code=1
D [04/Sep/2009:14:07:18 -0700] ReadClient: 6 POST / HTTP/1.1
D [04/Sep/2009:14:07:18 -0700] ProcessIPPRequest: 6 status_code=1
D [04/Sep/2009:14:07:18 -0700] ReadClient: 6 POST / HTTP/1.1
D [04/Sep/2009:14:07:18 -0700] ProcessIPPRequest: 6 status_code=406
D [04/Sep/2009:14:07:18 -0700] AcceptClient: 7 from localhost:631.
D [04/Sep/2009:14:07:18 -0700] CloseClient: 6
D [04/Sep/2009:14:07:18 -0700] ReadClient: 7 POST / HTTP/1.1
D [04/Sep/2009:14:07:18 -0700] ProcessIPPRequest: 7 status_code=0
D [04/Sep/2009:14:07:18 -0700] CloseClient: 7

The result of all this is that no job is spooled for this queue. Both lpq
and lpstat show no new job spooled, and the /var/spool/cups directory has
no new files added to it. The printer is disabled so we don't waste paper
while I'm testing, but the spool is accepting jobs:

# /usr/sbin/lpc status a65-laserjet
a65-laserjet:
        printer is on device 'socket' speed -1
        queuing is enabled
        printing is disabled
        no entries
        daemon present

This is confirmed by the fact that I can still spool jobs to this queue via
cups-lpd from a Mac.

The odd-looking message, "Error while reading file - Success", comes from
cups-lpd.c (line 714, in the source for v1.1.23). It would seem to indicate
that the previous fread or write operation returned <1 byte, though the
operation itself did not fail (meaning, errno was set to 0). I don't know what
that implies about the file cups-lpd got from the PC.

I've tried both
      server_args = -o document-format=application/vnd.cups-raw
and   server_args = -o document-format=application/octet-stream
in the /etc/xinetd.d/cups-lpd file, with no change in behavior. And I've tried
setting group=lp and group=sys in that file, also to no effect.

Can someone shed some light on this?