I have a pair of machines providing each other with nfs services.
One was rebooted, and did not reexport filesystems, nor did it start
mountd or nfsd.  (Not a problem, as I didn't need them any more.)

Unfortunately, this made find, df, and umount all hang *permanently* on
the other machine.  (Well, find from 12 hours ago in /etc/daily was still
in D according to ps...)... They did *not* time out.

They continued "correctly" as soon as I ran nfsd -tun 4 on the ex-server.

Shouldn't nfs filesystems time out and get errors after some amount of time?

-s