Are you running a kernel with DEBUG and DIAGNOSTIC - it sounds
	like whatever the problem is is corrupting kernel data, and
	DEBUG and DIAGNOSTIC may help to trac it down at the point of
	corruption, rather than sometime later.

		David/absolute

On Tue, 16 Nov 1999, Brian Buhrow wrote:

> 	Hello folks.  I'm trying to keep a NetBSD/sparc 1.4.1+ system running on
> an 85MHZ Sparc 5 running.  I keep getting kernel faults and alignment
> errors.  This is a stalk 1.4.1 system, except that I pulled down the latest
> locore.s and ccd.c files and remade the kernel.  These changes did not help
> the problem.  The machine is running a 2-disk ccd striped across 2 9GB
> disks and is trying to run inn.  This problem has existed more or less
> since NetBSD 1.3, but is very pronounced under 1.4.1.  The machine is now
> in a state where it cannot stay up more than 6 hours.  In addition, it
> seems to nibble on its news spool by corrupting inodes, and otherwise
> sabotaging itself.  I have it spewing kernel core dumps, but they don't
> appear to be usable by gdb and dmesg won't read them.
> 	Any help aanyone could provide would be extremely helpful.
> -thanks
> -Brian
>