I have a Linux box (1) that had a NFS mount from another Linux box (2).

Box 2 was down for a few days. Now box 1 has several hung tasks waiting
for the NFS to return. Box 2 has returned but NFS refuses to want to
see the previously mounted partition.

Now I can't umount the NFS partition. umount -f says it is in use.
I can't kill the tasks that are hung waiting for the NFS.

Now.. to complicate things, I would rather not reboot the box 1.

Any ideas how to clear out all of hung tasks and restore the NFS connection
without rebooting box 1.

The "in use" part is a bit of a worry...

Can you close the processes that usually
access the nfs shares?

Cheers,
Tink

No, I can not kill about half dozen processes
that appear to be waiting for the NFS partition.

In the past I have had this happen before where
the remote NFS box went down.. then came back.. and
everything just continued right along.. no problem.

This time, the remote box was down for 5 days and when
it came back the client box will will not seem to
recognize that it came back. I can ping it .. etc
and mount other things from it but the existing partition
mount seems to be "stuck"

The hung tasks have a STATE of "D" and new tasks
that start that want to access the NFS partition will
hang. Such as.. if I do a 'df' it will hang when it
gets to the NFS partition.

I know a reboot would fix everything.. but I have some
other tasks running right now that I would rather not
have to kill and restart right now. So I was trying
to see if there was a way to restore the NFS mount
without a reboot.

Ah. You have a problem with the D processes. That's uninterruptible sleep. You can't kill them, that's what uninterruptible means. The only way I know to fix those is to reboot. My NFS server doesn't like them either.