jmarin@pyy.jmp.fi (Jukka Marin) writes:

> Situation:

> NetBSD 1.2 NFS server
> NetBSD 1.3 NFS client

> Client has been whining "nfs server not responding" for several days.
> If I try to umount the disk, umount hangs.  If I 'df', it hangs.
> Several 'daily' runs have locked up as well.

> If I use tcpdump to see the traffic between the server and the client,
> _no_ packets are being sent or received when I try to umount the disk.
> If I mount it at a different mount point, packets are sent and received
> and the mount succeeds.  I can also umount this mount if I want to.

We've been seeing a similar problem here as well. Occasionally a
machine will lose sight of an NFS server, and just about everything on
that machine will become locked, requiring us to whack the reset
button (not nice on machine used by >10 people). Usually this happens
for no apparent reason (the server isn't down, is on the same segment
of ethernet, and isn't at all busy), but it also happens when the
server does go down; when that happens, any machine that attempts to
get something from its disk before it comes back up gets hosed.

We're using amd, and I've been figuring that amd doesn't propogate the
low level equivalen of "NFS server xxx is ok" back up through its
little quasi-NFS servers. Basically, once they think the server is
down, they think it's down forever until you reboot.

When this is happening, I too have tried mounting the "inaccessible"
disk on a different mount point, and it works just fine.

I know this is all a little incoherent, but I'm just wondering if our
problems might be related, and/or if anyone has an idea what I should
look at to try and get this working properly.