On Thu, 21 Jun 2001, Richard Bayly wrote:

> Paul, I don't want to sound like I'm taking sides, but it is very hard to
> give an answer to a customer regarding a system abort without looking at a
> dump. You can give them a general idea, but that's usually about it. This
> problem has been attributed to MFIO cards before. It has also been seen in
> relation to the console being switched to the remote support modem on ldev
> 21 and not switched back, which has caused problems.

Point taken, the response center gave very little information and I was
not pleased that it went right to the field without any HP pocket
protector types getting involved at all.  I suppose if there is no dump
they have better things to do, but the dearth of information added to the
fact that other problems in the past (ones we *knew* were software and
ended up being software) were pronouced MFIO board problems by the local
service did not contribute to my confidence.

> I would suggest you check out you system logfiles in logtool and check for a
> lot of I/O errors on a device prior to the problem occurring. This might
> give you a clue as to what went wrong. A dump would be able to tell us what
> device the problem occurred on.

Logtool showed nothing but an operator inserting a tape twenty minutes
prior to the problem on this last abort, and just a login notification (if
I recall correctly) some time before the previous abort.

>
> You would not be the first (and want be the last) to screw up a dump.

I had assumed that it was like the 9000, which requires no special
attention from me in most cases.  Some messages on boot (saving
dumparea to disk??) gave me that impression, too.
Got the correct procedure now. :-)

Paul

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *