Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 13 Feb 1997 05:43:02 -0500 (EST)
From:      Peter Dufault <dufault@hda.com>
To:        spork@super-g.com (spork)
Cc:        shovey@buffnet.net, questions@freebsd.org
Subject:   Re: drive failure?
Message-ID:  <199702131043.FAA13891@hda.hda.com>
In-Reply-To: <Pine.BSF.3.95.970213025219.1338B-100000@super-g.inch.com> from spork at "Feb 13, 97 02:54:34 am"

next in thread | previous in thread | raw e-mail | index | archive | help
> Hmmm...  The Buslogic verify wants to "destroy all data on the disk".  I
> don't think I want that.  Anyone have any ideas on this or know of the
> utility mentioned?

1. Verify that AWRE and ARRE are set in mode page 1.  Use the mode
page editor in scsi(8).  This will permit the drive to reallocate
on write errors and RECOVERED read errors.  It won't do anything
for unrecovered read errors such as you have logged.

2. See that you have POST ERROR turned off in that same mode page
to turn off the RECOVERED ERROR messages.  I suspect the driver is
treating that as an error instead of an informational message,
though I haven't verified that.

3. Now you must write something to that bad spot on the disk and
the drive will reallocate the sector.  Note that the Adaptec utility
can't properly synthesize data that it can't read either, though
it may have a sophisticated recovery technique, and so the suggestion
to turn the Adaptec utility loose on the disk may or may not be a
good idea.

The info number in the error message is the disk block that can't
be read.  If that block is on a swap partition dd the partition
full of zeros when no swapping is enabled on that partition and be
done with it.  If the bad spot is in a file system try to figure
out what to do - you have either lost part of a file or some file
system metadata.

Your options are restoring the partition after writing something
to the block or writing arbitrary data to that block to force a
reallocation and hoping for the best.  One option suggested by Rod
in the past is to read that block in a loop hoping at some point
it will succeed and the drive will slip it - this is unlikely to
work, though.

Ask again if you can't either restore the partition or use "dd" to
write something to that block.

If you're willing and able and have your disk backed up, FIRST pick
up ftp://freefall.freebsd.org/pub/dufault/scsinew2.tgz build it
and run the "zones.wish" tk script.  This should warn you about an
improperly setup disk and display the bad spots on it.  Or it may
crash your system.

Peter

-- 
Peter Dufault (dufault@hda.com)   Realtime Machine Control and Simulation
HD Associates, Inc.               Voice: 508 433 6936



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199702131043.FAA13891>