> I see it as fairly serious bug that shutdown would not be able
> to reboot the system after it caused random number of stuff
> to happen.

It is a lose-lose situation. In the NFS case, you cannot know how long
the server will be inaccessible, so you have to decide either to drop
data (bad) or to wait indefinitely (also bad).

IMHO a global timeout is an okay feature but is not good for the default.

If anything, each rc.d script should supply an appropriate default timeout
for itself. Things that just kill processes can time out quickly, but
network services that involve remote synchronization should wait much
longer before giving up.

Todd Whitesel
toddpw @ best.com