Public bug reported:
Binary package hint: cupsys
We maintain a legacy application which runs under dosemu on Ubuntu
workstations and prints 'raw' jobs using 'lpr -l' as $_printer_command
in dosemu.conf, using a Deban Sarge cups server as destination.
Printing is raw because the legacy app includes printer commands that
look like PCL, but break hpgltops.
>From time to time, when cupsys is restarted an 'old' job, previously
transferred to the server, is 're-printed', but not as a raw job: it is
passed to hpgltops, which loops forever and redlines the CPU.
In cupsd.conf, PreserveJobFiles is left at its default, stated as 'no'.
Sample /etc/cups/error.log lines, from when the job was originally spooled (the
'problem job' is job #11682 in this case):
I [23/Oct/2006:05:41:28 +0930] Adding start banner page "none" to job 11682.
I [23/Oct/2006:05:41:28 +0930] Adding end banner page "none" to job 11682.
I [23/Oct/2006:05:41:28 +0930] Job 11682 queued on 'main_laser' by 'va-kyle'.
I [23/Oct/2006:05:41:28 +0930] Started backend /usr/lib/cups/backend/ipp (PID
26126) for job 11682.
N [23/Oct/2006:05:41:28 +0930] [Job 11682] : Print file accepted - job ID
142489.
I [23/Oct/2006:05:42:41 +0930] Adding start banner page "none" to job 11683.
I [23/Oct/2006:05:42:41 +0930] Adding end banner page "none" to job 11683.
I [23/Oct/2006:05:42:41 +0930] Job 11683 queued on 'main_laser' by 'va-kyle'.
I [23/Oct/2006:05:42:57 +0930] Adding start banner page "none" to job 11684.
I [23/Oct/2006:05:42:57 +0930] Adding end banner page "none" to job 11684.
I [23/Oct/2006:05:42:57 +0930] Job 11684 queued on 'main_laser' by 'va-kyle'.
I [23/Oct/2006:06:04:34 +0930] Adding start banner page "none" to job 11685.
I [23/Oct/2006:06:04:34 +0930] Adding end banner page "none" to job 11685.
I [23/Oct/2006:06:04:34 +0930] Job 11685 queued on 'main_line_4' by 'va-kyle'.
I [23/Oct/2006:06:04:34 +0930] Started backend /usr/lib/cups/backend/ipp (PID
27093) for job 11685.
N [23/Oct/2006:06:04:34 +0930] [Job 11685] : Print file accepted - job ID 142493
Note also that there is no indication of the disposition of the two
subsequent jobs, which makes me suspect a (slow?) race of some kind;
grepping the log files confirms that they 'vanish without trace': they
are no longer in the queue or the spool area, and server log files
suggest that they were not, in fact, printed.
And when CUPS is next restarted:
I [23/Oct/2006:07:40:02 +0930] Full reload complete.
I [23/Oct/2006:07:40:02 +0930] Started filter /usr/lib/cups/filter/hpgltops
(PID 31667) for job 11682.
I [23/Oct/2006:07:40:02 +0930] Started filter /usr/lib/cups/filter/pstops (PID
31668) for job 11682.
I [23/Oct/2006:07:40:02 +0930] Started filter /usr/lib/cups/filter/foomatic-rip
(PID 31669) for job 11682.
I [23/Oct/2006:07:40:02 +0930] Started backend /usr/lib/cups/backend/ipp (PID
31670) for job 11682.
I [23/Oct/2006:07:41:41 +0930] Adding start banner page "none" to job 11704.
I [23/Oct/2006:07:41:41 +0930] Adding end banner page "none" to job 11704.
I [23/Oct/2006:07:41:41 +0930] Job 11704 queued on 'main_line_4' by 'va-kyle'.
Hpgltops consumes all available CPU and never exits, effectively
blocking the print queue; while this is also a bug, we're not too fussed
about that angle, as the job is arguably not 'well formed'; however,
having a critical component of the print system this vulnerable is
obviously undesirable.
Regards,
John.
** Affects: cupsys (Ubuntu)
Importance: Undecided
Status: Unconfirmed
--
CUPS drops jobs, reprints 'old' jobs with wrong options after a restart
https://launchpad.net/bugs/67897
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
