Hi .. I've not had to use NFS under NetBSD until today, and I
decided to give it a go on two production AS1200s.  Both were
built from the 1.5.1 release but have 1.5.2 kernels.  I went
ahead and checked the source for nfsd, mountd and rpcbind and
there are no differences in the two distributions, so I think
I'm fine there.

What I did was, I setup my exports file with the five 
filesystems I wanted to mount remotely using -maproot=root
and -alldirs options.  I started up the daemons on both 
machines and manually mounted each volume on the client.  I
was only doing very minimal things with these file system (copying
files less than 2mb) when the client hung.  I went over and
checked the server and it was so dead that it wouldn't even
respond to break signals to drop into the debugger via the
terminal server.  The client was still there, but was forced
into limbo while waiting for the server to recover (god I hate
that!)

Since this was all done remote I hopped in my car and started
thinking what is so wrong with NFS that it would crap out so
easily.  I've seen NFS problems with linux, but it usually 
because portmapper (our rpcbind) dies.  It was pretty discouraging
to see a simple scenario like this go belly up in less than an
hour of very minimal usage.

In researching this, I came across a bug report for 1.5.1_alpha
(not DEC alpha) in which the nfsd error: netbsd: nfsd send error 55
will crop up and bite you.  This bug report made it sound as if it
were port specific, It can be found here:

http://mail-index.netbsd.org/netbsd-bugs/2001/03/03/0006.html

Anyone have any clues?