Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!brl-adm!brl-smoke!smoke!rbbb@RICE.EDU
From: rbbb@RICE.EDU (David Chase)
Newsgroups: net.unix-wizards
Subject: Continuous Winchester use
Message-ID: <3166@brl-smoke.ARPA>
Date: Tue, 19-Aug-86 19:49:59 EDT
Article-I.D.: brl-smok.3166
Posted: Tue Aug 19 19:49:59 1986
Date-Received: Thu, 21-Aug-86 21:05:01 EDT
Sender: news@brl-smoke.ARPA
Lines: 25

I believe the man from CDC told us to leave ours running whenever
possible.  After several years of continuous (heavy) use we did experience
something like "bitrot" on one of our drives; whatever it really was, it
was NOT gradual; in the space of a few days we went from our normal error
rate (1-20/day) to unusable (10-20% of I/Os failing).

We NEVER had bearing problems, not even once, in our Winchesters.  In our
removable packs, however, we had several bearing failures and a few head
crashes.  We blamed this on repeated cycling and contamination.

It is also the case that running Winchesters are quite resistant to
physical shock; when unloading one, we (the man from CDC and I) bashed a
running drive on a running system so hard that the side panel fell off
(actually, the new Winchester did all the bashing when it slid off the
pallet).  No errors at all.  A Hyperdrive in a Macintosh was also observed
to be running just fine after the table holding it up collapsed.  (Does
anyone know how strong those air bearings really are?)

It is, however, probably a good idea to cycle a drive down and up maybe
once a month if it performs any sort of a self-test on power-up.  Our
drives would slowly drift out of (electrical) spec, and our first symptom of
a problem was usually failure to pass the self-test when they were powered
up.

David