Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 17 Jun 1998 00:04:24 +0200
From:      Stefan Esser <se@FreeBSD.ORG>
To:        Sean Eric Fagan <sef@kithrup.com>, hardware@FreeBSD.ORG
Cc:        Stefan Esser <se@FreeBSD.ORG>
Subject:   Re: scsi disk question
Message-ID:  <19980617000424.50072@mi.uni-koeln.de>
In-Reply-To: <199806161747.KAA06801@kithrup.com>; from Sean Eric Fagan on Tue, Jun 16, 1998 at 10:47:33AM -0700
References:  <199806161747.KAA06801@kithrup.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 1998-06-16 10:47 -0700, Sean Eric Fagan <sef@kithrup.com> wrote:
> I posted about this to -stable this morning, but now it's a hardware question
> I think :).

Well, I don't know how long ago "this morning" has been
in your part of the world, but I wrote a reply to your
mail to -stable just a few minutes ago ;-)

> I've got an IBM DCAS 3216W (2G UW drive) as my root disk, sd0.  After a power
> failure, /var/news/history.pag contained a non-recoverable bad block -- the
> kernel would try four or five times to access it, and then would fail,
> resulting in an I/O failure.  This persisted after a reboot.

Yes, it surely is a bad sector, which must be written to,
to be recovered. But I'm quite convinced that the drive 
has not been physically damaged and need not be replaced.

> After I found the file with the bad block, I removed it, and recreated it;
> this went well, and, so far, the system has continued to function.

Great!

> My question is:  will the disk now ignore this bad block?  Normally, I'd
> assume it would (it being an intelligent, scsi disk with Read-Write Error
> Recovery enabled), but, well, it didn't before :(.

I guess that sector was just incompletely written when 
the power failed, and the next write will just create a
consistent sector (data + ECC) again. There is no need 
to re-map the sector, in that case.

> I'm currently planning on getting a new disk today, but I'd prefer not to if
> possible, obviously :).

Guess I can understand that ;-)

> Anyone know for sure?

*You* can easily find out what the current state is: Just 
dd the complete surface of the disk to /dev/null. If you get 
another read error, then the sector has not yet received new
contents and ECC. You may try to force a write to that sector
by filling each and every sector of the disk (dd /dev/zero
into a file as root). If this does not work, then you may try 
to write to each sector on the disk via the raw device (you 
can write back the original contents). Use a fixit floppy, if
necessary. (You could try booting to single-user mode and do
the dd from the raw device back to itself even for a root 
partition, I assume.) 

That should fix it.


Be sure to backup your data before you proceed, whatever you 
try !

Regards, STefan

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hardware" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980617000424.50072>