more nfs wierdness

Occasionally, my nfs fileserver goes nuts. rpc.nfsd eats up 99% of one cpu's cycles and I can't mount, umount or do anything to an nfs system (the shell just hangs waiting for the server to do something).
I can kill rpc.nfsd on the server, but when I restart it, it just goes back to eating up a CPU worth of cycles and doing nothing.
I can usually reboot to fix it, but I want to know how I can fix the problem, not the symptoms.
The server is a dual PPro 200 running Debian unstable updated today.

edit: Rebooting didn't do anything this time around.

Have you tried starting and stoping portmap when that happens? Did you hard or soft mount the NFS filesystem on the client side?

moorcito said:

Have you tried starting and stoping portmap when that happens? Did you hard or soft mount the NFS filesystem on the client side?

Click to expand...

I've tried stopping and starting portmap (stopping all other nfs daemons before I stop it and starting them afterwards), but no go. I've tried both soft and hard mounting on the client side, but neither changes anything.
I'll probably just end up trying Samba or Coda. I'll set up a Samba share for the windows boxes on the network as soon as I can find the time to plow through the documentation, so I'll probably end up going that route unless Coda is as simple as nfs.