On Sun, 9 Dec 2001, Richard Rauch wrote:


> (Somewhere, I know that there's an option so that I can kill
> processes waiting on NFS.  If I enable that, and use it, the only danger
> is that application data may be corrupted if the client is trying to write
> data, yes?)

There is a mount-option (-i?) for mount_nfs to enable "Interruptible
mount", which means you can interrupt a process locked on NFS I/O. Quite
nice, although as you mention if the interrupted process is trying to
write, the write won't magically finish just because you press ^C. ;)

For unknown reasons, this interruptible wait only works on I/O, not on
the mount itself; I've bumped into this on several occassions. If the
NFS server actually dies, you can _not_ umount -f it. Unmounting an NFS
fs works iff the NFS server is alive, no matter how interruptible
your mount is and how many -f you supply to umount[0].

[0] If this is not a bug, it'd be interesting to know why it was
    implemented this way. I know unmounting a desynched fs may well cause
    data corruption, but I reckon that's what the -f flag is for; to tell
    umount I know perfectly well what I am doing. What am I missing?

-- 
/ali
:wq