On Sat, Jul 29, 2000 at 12:15:44PM -0400, Andreas Wrede wrote:
> The machine is neither an NFS server nor client, so I guess the
> problem is not restricted to NFS. I see about 10 transmit aborts per
> hour, relatively evenly distributed. Loading the interface via 'ping
> -s 1500 -f' does not appear to change to pattern.

No, but with UDP NFS the problem is worse. With TCP, the retransmit will
change the packet's size, so the retransmit won't trigger the bug.
But with NFS over UDP, the retransmitted packet will be exactly the
same, so it will trigger the bug again.

--
Manuel Bouyer <bouyer@antioche.eu.org>
--