>Hi Greg,
>
>saw your message (below) on this list only today. About a month or so I
>asked the same thing - answers were in the range:
>
> - check cables
> - check termination
> - replace the board
>
>But I feel it is more a software problem.
>I have changed cables/terminators and one of my disks, but the
>problem remains.
>
>Did you get any response beside the few mails on the list on Apr 9th ?

All I know so far is that I can run my old 1.2G kernel on that system
(which is our news server and as such does ALOT of disk IO) and it doesnt
get any disk errors.  But if I boot the 1.3.1 system it usually starts
getting scsi errors within 5-10 minutes and within about 30 minutes will
go into an infinite loop printing ``SENSE'' over and over again.

The versions of the files that run well are:

/*	$NetBSD: asc.c,v 1.39 1997/06/16 03:46:29 jonathan Exp $	*/
/*	$NetBSD: asc_tc.c,v 1.5 1997/06/22 07:44:05 jonathan Exp $	*/
/*	$NetBSD: ascvar.h,v 1.1 1996/09/25 21:07:56 jonathan Exp $	*/

I've exchanged several mail messages with Michael Hitch on this issue, he
is continuing to investigate the problem.  The investigation process has
been hindered by the fact that if DEBUG is defined in asc.c it fails
immediately with a scsi timeout instead of failing in the way it does
without DEBUG defined.  Bizzare.  So I've never been able to send him
any DEBUG output for the failure I'm trying to eliminate.


						--Greg


>> Apr  9 08:03:44 lazy /netbsd: asc_get_status: cmdreg 11, fifo cnt 7
>> Apr  9 08:03:44 lazy /netbsd: rz0: Illegal request, blk 3452431
>> Apr  9 08:03:44 lazy /netbsd: ccd0: error 5 on component 0
>> Apr  9 08:03:52 lazy /netbsd: asc_get_status: cmdreg 11, fifo cnt 7
>> Apr  9 08:03:52 lazy /netbsd: rz0: Illegal request, blk 3336336
>> Apr  9 08:03:52 lazy /netbsd: ccd0: error 5 on component 0
>[...]
>
>> The system in questino is a 5000/200 with the following scsi disks:
>
>Thats exacly what I have (but with different disks).
>
>> What does this mean?  I'm quite certain that the drives themselves are OK.
>
>I was not sooooo sure, but now ...
>
>Andreas