HP3000-L Archives

December 1997, Week 4

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
WirtAtmar <[log in to unmask]>
Reply To:
WirtAtmar <[log in to unmask]>
Date:
Tue, 23 Dec 1997 23:05:58 EST
Content-Type:
text/plain
Parts/Attachments:
text/plain (54 lines)
Mark Bixby asks:

>  Is there any way to configure a 3000 so that in the event of a crash, it
>  will automatically perform a memory dump (a DAT tape is always mounted and
>  online on the machine in question) and then do a START RECOVERY?

Mark,

My answer to the first part of your question is: "I doubt it." A crash is a
crash -- and is therefore an undefined, unexpected, quite serious situation.
The computer has essentially declared, "I've got an unreconcilable problem. I
don't know what to do -- but I don't trust the data I've got. If I were to go
forward at the point, corruption could potentially result. The best thing that
I can do now is to shut down and let somebody analyze the problem."

I think most people would argue that under these conditions, you shouldn't
have automatic restart capability following a system abort. The situation
could only be made much worse by repeatedly trying to autorestart a machine
that has declared itself to be defective in some (generally) unknown manner.

However, the real reason that I'm writing is to suggest that keeping a DAT
tape loaded at all times is probably not good for the drive or the tape. There
are a number of DAT drive models available now (and I haven't taken them all
apart), so I can't be sure of my answer -- but in general, a DAT tape that has
been "loaded" means that the drive has read some prefix material on the tape
and is now at the load point. The load point is not declared by some external
mechanism, such as a piece of reflector material mounted on the tape itself,
but due to the information it has just read off of the tape.

What this means at a minimum is that the capstan drive roller is squashed up
against the drive spindle (with the tape in between). If the roller should sit
in this position for any length of time (days), it's bound to develop a flat
spot on one side of the roller -- and that will forever after cause an
irrecoverable jerkiness in the tape's motion -- and that alone would make the
tapes written by that drive unreadable by most other drives.

Secondly, unless the drives have gotten a lot smarter, once the tape is at the
load point, the read/write helical scan head must remain under constant
rotation. Two relatively negative things happen by keeping the scan head
rotating. The first is that the bearings that support the rotating head are
very tiny and of amazing precision, thus they have a limited life. Secondly,
the constantly rotating scan head is going to abrade the writable magnetic
surface off of the tape itself.

To prevent either of these things from happening in a regular VCR tape drive
(which is just a macro version of a DDS), the tape "unloads" itself if it has
been paused for more than a few minutes. What this means is that the tape is
retracted away from the helical scan head, the capstan roller unloads, and the
tape is drawn back into the cassette. I've never seen this happen on a DDS
drive, so I must presume that the tape stays "loaded," with the helical head
rotating and the capstan pinch roller pressed into position.

Wirt Atmar

ATOM RSS1 RSS2