Red Hat Bugzilla – Bug 134237
Possible to send CUPS into an endless loop because of no hostname sanitychecks
Last modified: 2007-11-30 17:10:50 EST
Two computers: one running FC2 (kyrre), the other running
fc3t2(localhost). localhost is sharing the printer "HP2000C", and
kyrre is sharing "HP930C".
kyrre is a resolvable dns-name on the local dns-server, which points
localhost is also resolvable - trough /etc/hosts. This machines IP is
What if we now hit "print test page" for the printer "HP2000C" on
"localhost.localdomain" on kyrre's web interface "localhost:631"? Cups
would send it to itself, and when reciving the job, send it to itself,
send it to itself... In 10 secounds i suddenly had 100 jobs in the que
for "HP2000C", and the cups web interface went very slow and
unresponsive. I dont want to know if this would go on for say, one
minute, before i shut down cups and cleared out /var/spool/cups...
Version-Release number of selected component (if applicable):
On "kyrre" (FC2):
Steps to Reproduce:
1.Have a computer named "localhost" export a printer
2.Try to print to it
Cups should have detected that the domain name is not possible to
resolve back to the correct ip which the broadcast package came from -
and when it is not possible, use the ip adress instead of hostname.
Ip should have precedence over hostname.
Yup, the HP2000C is the one also found in this bug:
not that the bugs have anything in comon, just trying to draw some
attention to that bug as well :)
Really I think this is a network misconfiguration issue. Anyway, I've
reported it upstream: