A client has a Debian box, running mostly Debian Etch but there is CUPS from Lenny and perhaps some other things as well (the newer CUPS was needed due to a newer printer that needed a newer HPLIP, which itself was a major issue because it wanted to upgrade CUPS and HPLIP’s script was hiding the “dpkg –reconfigure” dialog, and and and and it was all a headache, but that was months ago).
So today out of the blue (says the client), CUPS just stopped printing. When trying to manually start the service we got these errors and then CUPSD would just hang:
shell# /etc/init.d/cups start
Starting Common Unix Printing System: cupsdmodprobe: module lp not found.
modprobe: failed to load module lp
modprobe: module ppdev not found.
modprobe: failed to load module ppdev
Looking at dmesg, we see these drivers load with the kernel and shouldn’t need to be modprobed:
shell# dmesg |grep lp
lp: driver loaded but no devices found
dmesg |grep ppdev
ppdev: user-space parallel port driver
We commented out the modprobe lines in the startup script /etc/init.d/cups but the hang would still happen. We eventually pinned it down to the file /etc/default/cups which has these lines:
# Cups configure options# LOAD_LP_MODULE: enable/disable to load “lp” parallel printer driver module
LOAD_LP_MODULE=yes
By changing that last line to “=no” we were able to get CUPS started properly. This only took about two hours to pin down; why doesn’t CUPS have a –verbose switch??