> it when it isn't. Is this a problem w/ linux's 1k nfs? 

Umm, linux has handled long packets for a long time. There used to be
an xkernel(sunos) that was hacked to mount the root at 1k instead of
8k blocks, but it was dropped as no longer necessary.

The most likely problem is the mounted /dev/console (and other
devices) being wrong. Turning on debugging on the mountd, or running
tcpdump on the linux side and deciphering the requests (nfswatch might
be better, not sure) will also help some.

							_Mark_