FS#16107 - [cups] 1.4.0-2 only prints one job

Attached to Project: Arch Linux
Opened by Jakub Schmidtke (sjakub) - Wednesday, 09 September 2009, 05:31 GMT
Last edited by Andreas Radke (AndyRTR) - Saturday, 14 August 2010, 21:06 GMT
Task Type Bug Report
Category Upstream Bugs
Status Closed
Assigned To Tobias Powalowski (tpowa)
Andreas Radke (AndyRTR)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 6
Private No

Details

Description:
cups only prints one job. after that I need to restart cupsd to be able to print anything else.

I have cups-1.4.0-2 and usblp blacklisted (as mentioned here: http://bugs.archlinux.org/task/15998).
After the first job which prints fine, when I want to print anything my printer (Samsung ML-2240, using driver for ML-1640) blinks twice and nothing else happens.
The first job is printed perfectly. I have splix 2.0.0-1 driver.
There are no errors in the log, and everything seems fine (there are no processes eatign 100% of CPU),
there is just nothing printed out...
This task depends upon

Closed by  Andreas Radke (AndyRTR)
Saturday, 14 August 2010, 21:06 GMT
Reason for closing:  Works for me
Additional comments about closing:  initial issue seems fixed. other may open new issues.
Comment by Tobias Powalowski (tpowa) - Wednesday, 09 September 2009, 06:47 GMT
-2 contains a patch that should fix it, seems not to be bullet proof yet :(
Comment by Jakub Schmidtke (sjakub) - Wednesday, 09 September 2009, 06:56 GMT
Yes, I know. Maybe that patch only fixed the 100% CPU problem?
I have had a number of issues recently, most of them are probably related to usblp,
before I blacklisted it. And at some point I have seen an insane number of [defunct] processes
(usb, and some ps convertion processes), but since usblp is blacklisted, and cups is in 1.4.0-2 with
fresh configuration (deleted /etc/cups, reinstalled packages, reinstalled the printer) I haven't seen those
processes, or any usblp-related dmesg entries. If only it was able to print more than one job....

Any suggestions, how can I try to figure out what is going on?
Comment by Tobias Powalowski (tpowa) - Wednesday, 09 September 2009, 12:37 GMT
You also have latest udev installed?
udev-146-2?
Comment by Jakub Schmidtke (sjakub) - Wednesday, 09 September 2009, 17:14 GMT
cups 1.4.0-2
hal 0.5.13-2
kernel-headers 2.6.30.5-1
kernel26 2.6.30.5-1
kernel26-firmware 2.6.30-1
klibc-udev 141-3
libcups 1.4.0-2
splix 2.0.0-1
udev 146-2
Comment by Andreas Radke (AndyRTR) - Wednesday, 09 September 2009, 19:16 GMT
please try some patches from http://cvs.fedoraproject.org/viewvc/devel/cups/

cups-str3311.patch STR #3292 and 3293 look promising. good luck. at least these are all upstream issues you should discuss in the cups bugtracker!
Comment by Jakub Schmidtke (sjakub) - Friday, 11 September 2009, 05:23 GMT
Those patches don't help.
I have submitted a bug: http://www.cups.org/str.php?L3326
And in the meantime I have discovered that printing txt files between PDF files somehow fixes it for the next PDF job...
Comment by Jakub Schmidtke (sjakub) - Friday, 11 September 2009, 17:37 GMT
I got a response on CUPS' bugtracker:

"Closed w/o Resolution: Please contact your Linux distributor for this issue."
Comment by Tobias Powalowski (tpowa) - Sunday, 20 September 2009, 19:43 GMT
should work now with 1.4.1
Comment by Jakub Schmidtke (sjakub) - Monday, 21 September 2009, 07:02 GMT
I still have the same problem. One pdf job and it won't print anything else.
Comment by Tobias Powalowski (tpowa) - Monday, 21 September 2009, 18:53 GMT
I removed the printer and readded everything again, then i have no issues anymore.
Is a HP Colorlaserjet 3505 and all is working fine.
Comment by Matthew Gyurgyik (pyther) - Monday, 21 September 2009, 19:40 GMT
If you are still having issues, enable debug logging for cups and look at the logs.
Also as tpowa, said try to remove and re-add your printer.
Comment by Andreas Radke (AndyRTR) - Wednesday, 23 September 2009, 19:31 GMT
please try a real printer. cups-pdf is something different.
Comment by Jakub Schmidtke (sjakub) - Saturday, 10 October 2009, 00:30 GMT
And I am still having issues.
I have removed cups, and /etc/cups, reinstaleld cups and splix again, and added printer from scratch.
Cups still prints only one job. After that I need to turn off the printer and turn it back on.
Restarting cups doesn't help.
I will attach 2 logs (debug level).
First one is - I turned on the printer, started cups, printed one PDF file (one page), stopped cups.
Second one - I started cups, printed exactly the same PDF file, stopped cups.
When I removed time column, and changed 'Job 148' to 'Job 147', most of those files are the same.
Comment by Jakub Schmidtke (sjakub) - Saturday, 10 October 2009, 00:30 GMT
Log files
   error_log.ok (73 KiB)
   error_log.failed (73.1 KiB)
Comment by Jakub Schmidtke (sjakub) - Saturday, 10 October 2009, 00:34 GMT
Hm, both logs with time information removed, and with string "Job 149" (the failed job) changed to "Job 148" (the OK one) in the .failed log.
   error_log.ok (56.2 KiB)
   error_log.failed (56.3 KiB)
Comment by Börje Holmberg (linfan) - Saturday, 17 October 2009, 17:12 GMT
cups is completely unusable - printer stopped working completely. I will need to find the old repository to completely downgrade all of cups and libcups. Dunno if this happened with gnome upgrade or what is going on.

I have a Canon with turboprint drivers. The printer is hooked to the router and I use the ipp protocol.
Comment by Andreas Radke (AndyRTR) - Saturday, 17 October 2009, 17:46 GMT
Jakub, please try to rebuild cups-pdf package against new cups. Maybe this is needed to fix the pdf-printer (remove the pdf-printer and add it again if you haven't done this so far).

Börje, ask in our forum or mailing list for help. The major cups will update needs some interaction, start with the frontpage news.
Comment by Jakub Schmidtke (sjakub) - Sunday, 18 October 2009, 04:29 GMT
I am away for a week. I will try to do that once I am back. Thanks!
Comment by Börje Holmberg (linfan) - Sunday, 18 October 2009, 08:02 GMT
This is too hard. Printer is completely dead. Do I have to buy yet another printer. Could you please advice me what would be a idiot proof printer this time :( I tried to downgrade cups, libcups, ghostscript and foomatic-filters, but the fucking shit is just dead. It just makes me puke. Grrr.... Fuck cups.
Comment by Börje Holmberg (linfan) - Sunday, 18 October 2009, 20:25 GMT
I got it to work by attatching the printer to the usb port - but but it fails with ipp, socket and http thru the router, so it does not work as a network printer any more.
Comment by c b (cb474) - Tuesday, 20 October 2009, 06:06 GMT
I am having a similar problem, with a Samsung ML-1740, using the foomatic driver.

It will print once. Then with any other print jobs the cups printer manager says the job is completed, but it never prints. If I turn my printer on and off again, then it will print once, and then same thing.

I have cups 1.4.1 and I have udev 146-2. Kernel 2.6.31.

I followed the instructions for blacklisting usblp in rc.conf. I did "chmod 664 /dev/bus/usb/005/002" after looking up my printer bus and device number with lsusb. (Although it wasn't clear to me if this was necessary with udev 146-2). I deleted my printer from the cups print manger and readded it.

But I still have this problem that it will only print once and then says jobs are completed but doesn't print them.
Comment by Börje Holmberg (linfan) - Tuesday, 20 October 2009, 06:24 GMT
I got the printer to print now thru the router, by editing /etc/cups/cupsd.conf.
Comment by c b (cb474) - Tuesday, 20 October 2009, 06:26 GMT
Also, one strange addition element is that after the first print job, when I do a second one cups will wake my printer, but nothing prints (as I said until I turn the printer off and ond again).
Comment by c b (cb474) - Tuesday, 20 October 2009, 06:28 GMT
I'm using the usb port. Printing through the router isn't an option for me. This is a laptop and I need to be able to print in different places, where there aren't always routers availbe and my regular printer doesn't even have a network connection.
Comment by c b (cb474) - Tuesday, 20 October 2009, 06:44 GMT
I have also just discovered that if I merely unplug the usb cable from my computer and plug it back in, then I can print another job. But this has to be done between every print job.
Comment by Jens Pranaitis (jensp) - Tuesday, 20 October 2009, 19:50 GMT
I think I'm having the same issue, running CUPS 1.4.1 from [extra] with the splix drivers (proprietary Samsung drivers have the same issue), I also see the usb backend segfaulting in dmesg. I'll rebuild cups with debug information and will see if I can get a backtrace. Could take a while though as this is more or less a production system.
Comment by Andreas Radke (AndyRTR) - Wednesday, 21 October 2009, 16:18 GMT
It seems only a few devices keep crashing after the first print job. You should try to collect all informations and ask again upstream. I don't it's a distribution/packaging issue here.

Raise debug level any maybe try a generic driver ppd to see if it's only related to certain ppds. It could be even a gcc optmization bug. Try to rebuild cups packages with unset CLFAGS.
Comment by c b (cb474) - Friday, 23 October 2009, 08:09 GMT
Andreas, thanks for the suggestions. I did post a bug report upstream just now: http://www.cups.org/str.php?L3388+P0+S-2+C0+I0+E0+M20+Q

Unfortunately, everything you're suggesting is way over my head. Also the one thing that makes me think this bug may be specific to Arch is that when I search online I don't see any reports of it elsewhere, whereas I do see reports of the other issues people were having with upgrading to Cups 1.4 (e.g. the problem with usblp). I thought for sure at least in Ubuntu some people would be reporting it. Perhaps I just missed it somewhere.

I have strangely found now that with the foomatic driver only one job prints, the splix driver prints every other job (but shows all as completed successfully), and the unified driver from Samsung itself prints all jobs as expected (except the printing is way off center--printing is also off center with splix, only foomatic does a halfway decent job, which is why I want to use foomatic).

Comment by Matthew Gyurgyik (pyther) - Friday, 23 October 2009, 11:26 GMT
It probably isn't arch specific because as far as I know there are not many distros using cups 1.4. In fact I have a strange issue with my one printer where I will get 2-4 good prints and then it just prints garbage. I filed a bug report on the cups website and they told me to contact the manufacture (it worked just fine in 1.3.11).
Comment by c b (cb474) - Saturday, 24 October 2009, 01:39 GMT
Sheesh, they closed my bug report at CUPS saying that they don't support issues with foomatic drivers. Of course they didn't even read my bug report closely enough to see that it also happens with Splix. And I provided the link to this bug report which shows at least one person is also having the problem with the proprietary Samsung driver. And I pointed out that for everyone the problem began with the upgrade to CUPS 1.4.

Would people please post bug reports at CUPS so they get the message: http://www.cups.org/
Comment by Matthew Gyurgyik (pyther) - Saturday, 24 October 2009, 13:27 GMT
I feel your pain. They pretty much did the same thing to me. I posted a bug report and they completely ignored the fact it worked with cup 1.3 and it was with the raw queue! They told me to contact Samsung ;(
Comment by Jakub Schmidtke (sjakub) - Saturday, 24 October 2009, 16:49 GMT
And my bugreport was closed because they don't support:
- patched versions like the one in Arch (with the patch that was applied to svn, just not released yet)
- third party drivers (splix), even though it worked just fine with cups 1.3
Very apple-ish attitude :/
Comment by Marcos Alves Pintar (mapintar) - Saturday, 24 October 2009, 21:26 GMT
The svn version (1.4.2) work for me. Problem is solved.
I'm make a PKGBUILD
The other file user for make a package is the same of original PKGBUILD version 1.4.1, in extra (sorry for my english).
   PKGBUILD (3.3 KiB)
Comment by Jakub Schmidtke (sjakub) - Sunday, 25 October 2009, 01:35 GMT
svn cups doesn't solve the problem for me. Reconnecting the USB cable works.
Comment by Marcos Alves Pintar (mapintar) - Sunday, 25 October 2009, 02:06 GMT
Try remove packages foomatic-db, foomatic-db-engine and foomatic-filters and recompile svn version. These packages is not necessary for splix and maybe the error is caused by them.
Comment by c b (cb474) - Sunday, 25 October 2009, 08:38 GMT
Are people finding this to be true, that foomatic causes the problem? I really prefer foomatic to splix, because splix prints everything off center for me.
Comment by Daniele Marinello (marinz) - Sunday, 25 October 2009, 09:46 GMT
@http://bugs.archlinux.org/task/16107#comment51884 :

I can't add printer with cups 1-4-2 svn! it displayed an error message
Comment by Marcos Alves Pintar (mapintar) - Sunday, 25 October 2009, 21:13 GMT
Sorry for all. I'm using a script for remove module ohci_hcd when job is send for the printer, but put in /etc/rc.local and forgot.
The script is not elegant, but work.
In my system, usb printer is a second file in /dev/bus/usb/002. Adjust for your.
   notify.py (1.4 KiB)
Comment by c b (cb474) - Sunday, 25 October 2009, 21:19 GMT
Thanks for the script. Sorry to as a very simply question. Should I copy the entire text of the script into /etc/rc.local? Or do a put the script somewhere else and then put some sort of line in /etc/rc.local to execute the script? If you have a chance to provide a step by step explanation, I'd be grateful. Thanks.
Comment by Marcos Alves Pintar (mapintar) - Sunday, 25 October 2009, 22:39 GMT
Put the script in /usr/local/bin and add this line in file /etc/rc.local:

python /usr/local/bin/notify.py

Restart the system and test if work. No other action is necessary.
Comment by c b (cb474) - Monday, 26 October 2009, 07:04 GMT
Thanks for the quick reply and the instructions. So when I did this, my system would hang on restart. After it was done loading all the daemons, it would just hang. I had to manually power off my computer and boot off a different partition (that I have antiX on), so I could comment out the "python /usr/local/bin/notify.py" line in /etc/rc.local. Then I could restart, but of course it seems I can't use the script.
Comment by Marcos Alves Pintar (mapintar) - Monday, 26 October 2009, 12:56 GMT
Add & in line, like this:

python /usr/local/bin/notify.py &

You can too execute the script in terminal, after system load.
Comment by Jim Needham (jame5ie) - Tuesday, 27 October 2009, 07:50 GMT
I'm seeing the same issue (cups 1.4.1-1, udev 146-2, Samsung ML-1610, Splix 2.0.0-1 driver. I don't have foomatic installed). Printing fails after successfully printing a few documents (usually around 1 to 4) from openoffice, or evince (haven't tried printing from anything else). There is activity on the printer light and cups reports the job as completed but nothing is printed. If I wait a couple of minutes and try again printing will work (power cycling the printer also works). There is also another issue where nothing is printed and cups reports the printer as being in the paused state -- in that case if I manually resume the printer from cups admin pages the job prints, but I would expect it to resume automatically.
Comment by Andreas Radke (AndyRTR) - Tuesday, 27 October 2009, 12:00 GMT
splix is not supported! neither in Arch official repos nor by cups! I'll close this bug instantly if you cannot confirm the issue with foomatic/gutenprint drivers.
Comment by Bogdan Szczurek (thebodzio) - Tuesday, 27 October 2009, 22:43 GMT
I'll test the issue tomorrow with foomatic (I'm using splix right now having the same SNAFU as many people here)—will report back.
Comment by Matthew Gyurgyik (pyther) - Tuesday, 27 October 2009, 22:53 GMT
My problem is slightly different. I can print 2-3 pages before I get garbage. This happens with foomatic, hpijs-plc5e, splix, and even with the raw print queue. At least with my issue I believe it is a underlying problem with the new usb subsystem, as I get the same problem with the raw print queue. Not sure if this is enough or similar to keep this report open.
Comment by c b (cb474) - Wednesday, 28 October 2009, 07:57 GMT
I already reported above that I have this problem with the foomatic driver. Also Börje Holmberg seemed to indicate above that he was having the problem with the foomatic driver.

Also, doesn't the fact that people have had this problem, upon upgrade to CUPS 1.4, with Foomatic, Splix, and the proprietary Samsung driver, indicate that this is something more than just a driver issue and perhaps a problem with CUPS itself? Doesn't that make the information about Splix and the proprietary drivers relevant?
Comment by Andreas Radke (AndyRTR) - Wednesday, 28 October 2009, 08:37 GMT
This doesn't seem to be the initial bug that was fixed upstream. Please open new bugs for remaining issues to follow upstream discussions and solutions. Try to find out if only certain drivers are affected. So far I cannot see a general packaging issue because it works for most users now. Also please have a look what Fedora is backporting/patching in their devel packages http://cvs.fedoraproject.org/viewvc/rpms/cups/devel/ .
Comment by c b (cb474) - Monday, 08 February 2010, 23:54 GMT
I continue to have exactly the problem described in this bug report. It was never fixed for me. I'm using cups 1.4.2-3. I've tried in both x86_64 and i686 and get the same result. What version is it supposed to be fixed in? I never understood why this was closed I tried all the suggestions and every different possible driver for my Samsung printer and I continue to have exactly the problem this bug report describes.
Comment by Jan de Groot (JGC) - Monday, 08 February 2010, 23:56 GMT
Thomas showed me some printouts made with his samsung printer using the pxlmono driver. The first job is fine, every second job prints a lot of crap and then prints the page. Somehow this printer will buffer crap from the first task and take it with the next printing job. I don't know what printer Thomas has exactly and what of it is implemented in hardware, but if your printer is a pure GDI printer, it will choke in the crap left from the first task and not print anything at all.
Comment by c b (cb474) - Tuesday, 09 February 2010, 05:39 GMT
Thanks for reopening the bug report. I think I spoke a little too soon. Sorry about that. The splix driver, which was not working for me before, does not now have this problem. Although splix prints off-center, which is annoying and makes it not entirely useful. The foomatic/gdi driver does continue to have the problem of failing to print second jobs, until I unplug and replug the usb cable. I'm using a Samsung ML-1740.
Comment by Jakub Schmidtke (sjakub) - Tuesday, 09 February 2010, 12:38 GMT
My splix prints fine, but still only one pdf job. With txt files it is ok. Also, printing a txt file after pdf job fails, but somehow re-enables the printer and next job (either txt or pdf) works fine (and if the second one is pdf job it blocks again after that).
Comment by Marcos Alves Pintar (mapintar) - Tuesday, 09 February 2010, 13:58 GMT
I had this problem for several months. It's caused by blacklisted usblp.
Comment by Francis Carroll (fdac) - Wednesday, 17 February 2010, 11:51 GMT
Same problem here with Samsung ML-2010, splix 2.0.0-2, cups 1.4.2-3, udev 151-3
Can print one job, usually a test page, then all subsequent jobs show up in cups job list as 'complete' but nothing is actually printed.
Comment by Andreas Radke (AndyRTR) - Thursday, 18 February 2010, 21:42 GMT
you should do some talking to the upstream devs. we just ship what is recommended upstream. if some printers still don't work properly they should fix it.
Comment by Andreas Radke (AndyRTR) - Wednesday, 31 March 2010, 06:51 GMT
can you please check if all drivers(foomatic+splix) have the job issue? please check also cups error.log with debug level set to debug2.
Comment by Bogdan Szczurek (thebodzio) - Wednesday, 31 March 2010, 19:37 GMT
I've got similar problem with raw device (printing to Samsung ML-2010PR via samba). E.g. job is sent, shows as completed but actually doesn't give any printout. Next the same document is sent to printer again (another print op from windows client) and this time it all works as it should. Could someone confirm? I'm using cups on print server at work so this thing is rather crucial for me ;).
Comment by Andreas Radke (AndyRTR) - Saturday, 10 April 2010, 10:34 GMT
Only a few printer drivers seem affected. Affected users should contact cups/foomatic/splix and ghostscript upstream developers to localise the problem. I cannot do much for you.
Comment by Robert (Bysshe) - Monday, 12 April 2010, 18:55 GMT
I have done everything including the udev rule. Problem persists in the Splix driver. System fully upgraded. Will report bug to Splix.
Comment by Ondrej Podhajsky (podhaj) - Tuesday, 13 April 2010, 00:21 GMT
I got same problem (thebodzio) with Samsung ML 1640. Will report Splix bug.
Comment by Andreas Radke (AndyRTR) - Sunday, 02 May 2010, 12:23 GMT
check this out: http://mailman.archlinux.org/pipermail/arch-general/2010-April/012867.html
Comment by Jakub Schmidtke (sjakub) - Sunday, 02 May 2010, 17:47 GMT
I have tried this patch and I am still unable to print more than one pdf job at a time.
Comment by Jakub Schmidtke (sjakub) - Sunday, 02 May 2010, 17:47 GMT
(Modified PKGBUILD: http://pastebin.com/0rZtsb60 )
Comment by Andreas Radke (AndyRTR) - Wednesday, 12 May 2010, 21:14 GMT
Any response from the splix guys?
Comment by pablo (pablo) - Saturday, 19 June 2010, 04:51 GMT
After patching cups have you taken out usblp from the blacklist?
Comment by Andreas Radke (AndyRTR) - Saturday, 14 August 2010, 21:05 GMT
several issues seem to be mixed here. most are probably solved. closing this one. whoever is still affected please open a new issue and please also contact upstream devs/file upstream bugs.