https://bugzilla.novell.com/show_bug.cgi?id=166912 ------- Comment #15 from schuetzm@gmx.net 2007-01-23 11:35 MST ------- (In reply to comment #14)
I'll not modify fuser(1) by any change within its code, this because any change can break also the expected common behaviour in other cases with and without nfs mounted partition ... beside this, do you have tested the settings I've mentioned?
I will test it tomorrow at work.
IMHO the bash function nfscheck() will list all nfs mount points in ALLMTP and the `fuser -k -m' will kill the program accessing this mount point
I actually checked the source of fuser; AFAICS it stat's the mount points unconditionally, before it does anything else (like sending signals), but only uses the information if -m is given. Deferring the stat until it is actually needed would not change the behaviour, but it would only help with the alsasound script, which doesn't use -m. For ifdown-connection, where -m is needed, another solution would be necessary.
... if this really does not work then your mentioned `umount -l' may help if it would be done before any the fuser call its self.
Probably, but in this case, fuser probably wouldn't find the processes with open files on the now detached mount point, and consequently wouldn't kill them (haven't tested it though). Note that I tested umount -l in alsasound, and it didn't work there (see comment #5).
Nevertheless, changing a common tool like fuser without knowing all side effects is not an option for me ;)
I understand that.
Maybe Christian knows more about ... Christian?
Beside this NFS mounting on /mnt is not very common therefore this is IMHO not a major bug, to be exact, this is AFAIK the first report about.
I think you misunderstood that... /mnt was only an example. It is of course the same on all other mount points. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.