Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site natmlab.OZ
Path: utzoo!linus!decvax!mulga!munnari!natmlab!ronb
From: ronb@natmlab.OZ (Ron Baxter)
Newsgroups: net.unix-wizards
Subject: Re: more tales of RA81s--handling of bad sectors
Message-ID: <71@natmlab.OZ>
Date: Fri, 23-Nov-84 03:47:01 EST
Article-I.D.: natmlab.71
Posted: Fri Nov 23 03:47:01 1984
Date-Received: Sat, 10-Nov-84 21:29:01 EST
References: <bbncca.1111>
Reply-To: ronb@natmlab.OZ (Ron Baxter)
Organization: CSIRO Nat. Measurement Lab. (Maths+Stats & Applied Phys), Sydney,Australia.
Lines: 21

In article <bbncca.1111> sdyer@bbncca.ARPA (Steve Dyer) writes:
>.......  Within the last few days, two of the drives have developed
>"hard errors" which were not present at the installation.  Naturally a few
>of them reside in the swap area, thus randomly killing processes, and a few
>reside in files like /usr/lib/aliases.pag.  Only a minor headache!

Some months back one of our RA81s developed bad-blocks. I had assumed
that RA81s were immune from bad-blocks due to their intelligence.  Our
Field Engineer said that while the RA81 would not write on a bad-block
(ie a block that does not read-check after writing), it has no special
magic to cope with blocks in an existing file that were "good" and then
go "bad".  His advice was to dump the whole file-system if possible (it
wasn't really), and then restore from backup and the bad-blocks should
go away.  THEY DID!.  So I do not understand how "bad-blocks" in the
swap area could occur, while bad-blocks in an aliases file are easier
to understand.

PS it turned out later that the appearance of "bad-blocks" on our RA81
seemed to ba associated with the gradual failure of a power supply (the
voltages were just going too low).  Besides bad blocks this problem also
made the drive go off-line by itself.