Last night was the first time we stressed our array of NetBSD batch
processors since upgrading from 1.2.1 to 1.2G Snapshot.

By this morning two of them had paniced with "lockmgr: locking against
myself".

The traceback looks the same as this pr.

Machines are PentiumPro 200 with 128Mbytes memory, SMG 100baseT
network card.  They were running one compute/memory/NFS intensive
program. From the log it looks as if the machines died soon after one
of our major programs fired up and would typically be doing heavy
paging whilst sorting out its memory allocation at this point.

I have left both machines I the Debugger in case I can be pointed in
the right direction to gain any useful information from them.

Tim.