Bugzilla – Bug 309132
avahi-daemon and avahi-dnsconfd claims to be dead after boot.
Last modified: 2007-09-13 21:03:12 UTC
I have installed my computer and chacked these daemons if they are running: /etc/init.d/avahi-daemon status Checking for Avahi daemon: dead /etc/init.d/avahi-dnsconfd status Checking for Avahi DNS Configuration daemon: dead but they are not running.
jpr@rook:~> /etc/init.d/avahi-daemon status Checking for Avahi daemon: Failed to open PID file: Permission denied dead jpr@rook:~> /etc/init.d/avahi-daemon status Checking for Avahi daemon: Failed to open PID file: Permission denied dead Its not actually dead. The pid file is just not accessible because avahia-daemon via libdaemon tries to open it read/write instead of just read as a regular user when passed the -c option.
Lowering severity, since the daemons aren't actually dead.
From lennart upstream: If I remember correctly I open it as RDWR there because of a limitation of the file locking APIs on some OSes, where locking files you don't have opened as O_RDWR doesn't work. However, I agree that the bug report mentioned is worth fixing. So it might be advisable to add some code to reopen the file as O_RDONLY if O_RDWR fails, and also to try F_RDLCK as a fallback if F_WRLCK fails. That should be reasonably safe even on more exotic OSes. If have prepared a patch for this in SVN r131 (although I didn't test it) If this works for you I will ship this in the next release of
Mauro can you get this patch in before the end of this week?
The patch is actually for libdaemon FYI.
Created attachment 163991 [details] Patch for this bug, according to jpr's comment.
Applied patch suggested in comment #4. Submitted and tested --for compilation-- via abuild & mbuild. In sake of completness, the patch was added to the bug. Test packages available here: http://w3.suse.de/~mauro/bnc-309192/
The right url is: http://w3.suse.de/~mauro/bnc-309132/
Created attachment 163993 [details] Patch for this bug, according to jpr's comment.