Description of problem: I think since upgrading cups and/or related packages in a recent "yum update", I noticed that cups is flooding its access_log with lines such as the following at a rate of about 5000 lines/minute: localhost - - [11/Sep/2006:12:15:24 -0600] "POST / HTTP/1.1" 200 129 CUPS-Get-Printers successful-ok As you can imagine, this will (and did) fill up a 2-3 GB filesystem fairly quickly. I'm running cups 1.2.3 on FC5 with a local usb printer configured and working. I'm also using samba 3.0.23a and had suspected that my observed cups problem may be caused by interactions with samba (because whenever I stop cups during these log floods, I see an smbd write failure in the system logs). However, I found a reference on the CUPS site that indicates that the problem is actually due to eggcups (rpm desktop-printing). I have tried using both the default cupsd.conf configuration file that is distributed with the latest rpm as well as the previous version that I was successfully using on an earlier FC4 system. A recent search turned up the following reported related bug fix for debian: http://svn.debian.org/wsvn/pkg-cups...n/?rev=154&sc=1 Another related posting was found on the CUPS site: http://cups.org/newsgroups.php?s1+gcups.bugs+v4+T0+Qaccess_log+flood Version-Release number of selected component (if applicable): desktop-printing-0.19-6 cups-1.2.2-1.8 hal-cups-utils-0.5.5-1.2.fc5.2 samba-3.0.23a-1.fc5.1 How reproducible: Very Steps to Reproduce: 1. tail -f /var/log/cups/access_log & 2. service cups start 3. wait a few minutes and watch the resulting flood of log entries Actual results: The /var/log/cups/access_log file starts growing very rapidly. Expected results: The log file shouldn't grow so rapidly (~ 10K/sec or .75 GB/day). Additional info:
This: http://cups.org/newsgroups.php?s1+gcups.bugs+v4+T0+Qaccess_log+flood is not related. It refers to the polling mode of eggcups, which is a fallback used only when D-BUS is unavailable. This polling mode does not cause 5000 lines/minute logging. This URL: http://svn.debian.org/wsvn/pkg-cups...n/?rev=154&sc=1 seems incomplete, so I cannot comment on it. Looks like this bug is a duplicate of another one that has been reported (bug #205619), but just to be sure: do you see high CPU load when this occurs?
I have more information. In fact, I seem to have solved my problem. First, though, to answer your question: yes, I was seeing high CPU load (the cups process was hogging around 90% of the CPU) when this occured, and I needed to stop the cups service in order to free up the CPU and stop the logging, of course. Also, certain client commands like "lpq" or the "Default Print" applet (/usr/bin/gnome-default-printer) would hang and need to be killed. Now, what I did to set things right: I removed my previous PRD file(s) from /etc/cups/ppd and cleared out the /etc/cups/printers.conf and then re-added/configured my printer using the web-based admin interface (http://localhost:631). I also restored my cupsd.conf file to match the default version that comes with the rpm distribution, though the various cupsd.conf settings tried were never the cause of the problem. I'm fairly certain that the previous PRD file was the culprit. I had upgraded systems from FC4 and at the same time upgraded hardware (though kept the same architecture). The printer is the same, but it is now connected over USB instead of the parallel port. Since fixing the PRD, the access_log now shows only one record every 5 minutes or so (not consistently, but varying from maybe 3-10 minutes). I assume this is due to the normal behavior of samba and/or eggcups, etc. Printing is working fine both locally and over samba (ie. from remote Windows clients). So, all is well now. (Of course, it's probably still worth looking into the prevention of run-away log floods -- even due to misconfigurations.)
Sorry, one last side-note: the reason I didn't immediately think to replace the PRD file is that the printer was actually working fine for some time after setting up my FC5 system -- I had been printing successfully both locally and over samba. Then on 9/7/06, when I last ran a "yum update" and upgraded cups to version 1.2.3, I started noticing the problem with the cups access_log filling up my /var filesystem.
This was a known problem in CUPS-1.2.3 and has been fixed since then. *** This bug has been marked as a duplicate of 205619 ***