on 6/14/00 11:16 PM, NetBSD Bob at nbsdbob@weedcon1.cropsci.ncsu.edu wrote
something like:

> I seem to be getting some 5000/200 board errors that cause it to either
> abort or lock up.  It originally was a scsi disconnect error, but the last
> one I saw, on Ultrix, indicated a buffer chip malfunctioning.  Anyone know
> anything about chips going bellyup on 5000/200 machines?  Is such a problem
> endemic to the things?  I have heard one or two others locally suggest that
> scsi busses on these are fragile, and maybe I am running into something like
> that....  Any insights are appreciated.

    It's not terribly likely, DECstations are built like tanks. I would lean
toward there being a problem with your cables or connector than a problem
with the chip itself. This error message could be the result of an
intermittent open in a SCSI cable. AFAIK the SCSI buffer is on chip. (inside
the 53c9x) I'd try reducing your SCSI chain to a single known good drive,
enable terminators and termination power and try swapping cables around. If
it still does it, it is likely to be the motherboard.
 
> NetBSD-1.4.2 or 1.4P or 1.4T seemed a little more sensitive to the problem
> than Ultrix was, although Ultrix may have just let it go until it got real
> bad.

    Well try 1.4.1 and see if it gives you problems too (compile a custom
kernel with verbose scsi errors would be a good idea too). If 1.4.1 gives
you trouble then it's definitely the motherboard/SCSI chain rather than a
code problem. You might also try t 5/ to list the PROM device tests, and run
sdiag with your target attached. t/5/sdiag 1 for ID 1 IIRC.

> Anyone else have 5000/200 scsi or mother board problems?

    I'm not running mine anymore, I hacked up the case to put a 240
motherboard inside. I have a spare 5k/200 motherboard in my closet that
isn't doing anything if you want it. It runs hot, but it works.


    Chris