On 10 December 2010 10:09, sransome simon@nosher.net wrote:
Jenny Hopkins wrote:
Hi ALUG,
I have an annoying CUPS printer problem:
Does anybody know if this means the printer itself hasn't enough memory, or if the cups server has a file limit set somewhere? Or any other ideas?
There appear to be several limits configurable in CUPS, for instance, from "man cupsd.conf":
LimitRequestBody Specifies the maximum size of any print job request.
or from http://www.cups.org/doc-1.1/sam.html:
"The/ job-k-limit/,/ job-page-limit/, and/ job-quota-peiod/ options determine whether and how quotas are enforced for a printer. The/ job-quota-period/ option determines the time interval for quota tracking. The interval is expressed in seconds, so a day is 86,400, a week is 604,800 and a month is 2,592,000 seconds. The/ job-k-limit/ option specifies the job size limit in killobytes. The/ job-page-limit/ option specifies the number of pages limit."
(these are references to different versions - 1.3.10 and 1.1 respectively, so may now be one and the same)
So you could furtle around for those, or try setting them to arbitrarily large numbers (I think there's an "Umlimited" option). I had a very similar problem on my previous server (an ancient Pentium Celery with only 750Mb memory), which would fail to queue large photo print jobs to an Epson R800. It was a while ago, but I do remember also coming across some OS/Kernel limits on print job size too. However, tweaking these didn't seem to resolve the problem, and I can't remember what they were anyway - sorry.
However, as things are breaking rather than erroring in a sensible way, then it's perhaps just as likely that the printer simply has insufficient memory to render the entire job - a 750kb PDF could be several megabytes by the time it's been turned in to page data, all of which generally happens in [printer] memory before printing starts. Broken pipes generally occur when the receiving end of a data connection (the pipe) unexpectedly dies or breaks off the session, so this would imply that the printer is just falling over or is simply deciding it's not going to accept any more data as its buffer is full.
Not much help otherwise, I'm afraid ;-)
This is very helpful, thanks - it spurred me on to further tests, where I found the same document will print from a debian machine to the Xerox, eliminating the printer being at fault. I will snurfle about the ubuntu setup trying to spot the differences in cups settings.
Many thanks,
Jenny