In October of 1996(!), PR kern/2797 was filed.  It appears to still
be an issue (though the PR was closed in 1999).  I've seen the
issue come up from time to time, but don't remember ever seeing a
fix.

The problem is if you have an NFS-mounted filesystem, and then lose
the network connection, it appears to be impossible to break the
mount.  Your only options are to restore the link or throw the power
switch

On a recently installed NetBSD box, this bug has reared its head again.

I hoped that if I let it sit long enough, a timeout would pass and it
would complete anyway.  I let it wait a couple of hours, which seems
more than adequate for any such timeout.  (^&


The behavior seems to match that of the 1996 PR.

Can someone confirm?

If so, should a new PR be filed, or can the old one be re-opened?
(Or maybe, seeing how it was fixed before, and what has happened
to the files since, perhaps it can be fixed in short order this
time around?)

In about half an hour, I'll see if reconnecting the old NFS server
to the network will let the client finish its hours-long umount.

-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/