Subject: Re: help interpreting SCSI errors
To: NetBSD Users <netbsd-users@NetBSD.org>
From: Greg A. Woods <woods@planix.com>
List: netbsd-users
Date: 05/06/2007 11:25:29
--pgp-sign-Multipart_Sun_May__6_11:25:27_2007-1
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

At Thu, 3 May 2007 09:11:26 -0700, Bill Stouder-Studenmund wrote:
Subject: Re: help interpreting SCSI errors
>=20
> On Wed, May 02, 2007 at 09:56:39PM -0700, Aaron J. Grier wrote:
> > On Fri, Apr 27, 2007 at 12:42:18PM -0400, Louis Guillaume wrote:
> > > Can anyone help interpret this message?
> > >=20
> > > sd1(esiop1:0:0:0):  Check Condition on CDB: 0x28 00 02 6f 65 83 00 00=
 04 00
>                                                 aaaa bb ccccccccccc dd ee=
eee ff
> > >     SENSE KEY:  Media Error
> > >    INFO FIELD:  40854918
> > >      ASC/ASCQ:  Read Retries Exhausted
> > >          SKSV:  Actual Retry Count: 63
> > >=20
> > > raid1: IO Error.  Marking /dev/sd1a as failed.
> >=20
> > > I know it means that netbsd had problems writing to this drive, but how
> > > can I research to figure out if there's something wrong with the drive
> > > itself?
> >=20
> > if you can find the vendor's manual, you can sometimes get extra
> > information by decoding the error condition (CDB: 0x28 ....) .
>=20
> There's not that much more to decode.

It would be nice if the rest were decoded, with decent field labels.  :-)

Someone sent me some code that does try to decode all the magic in a CDB
string once upon a time....


> Sounds like a good idea. It might also help to write something to that
> sector. I have heard it bounced around on the lists that a lot of drives
> remap sectors only on write. So maybe if you put something there the drive
> can fix itself.

I'm not sure of the latest drives, but many drives don't have the AWRE
and ARRE bits set by default in their mode page 0x01.  I.e. they won't
automatically re-allocate bad sectors until those bits are turned on.

Unfortunately and sadly NetBSD has never had a native tool to edit drive
mode pages.  I've always resorted to building the ancient FreeBSD "scsi"
tool.  I think it still works OK on the most modern releases, though the
last time I had cause to build it I was on NetBSD/sparc 1.6.2 and it
worked great.

I proposed adding this "scsi" tool to NetBSD once a very long time ago
(back when it was still maintained in FreeBSD) but someone in their
infinite wisdom thought it was better to have no mode-page editing tool
at all rather than to have one that was imperfect and not designed to
exceed the highest standards possible.  Today it probably wouldn't take
too much coding effort to roll the mode page editing code from the old
"scsi" code into scsictl(8).

--=20
						Greg A. Woods

H:+1 416 218-0098 W:+1 416 489-5852 x122 VE3TCP RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>       Secrets of the Weird <woods@weird.com>

--pgp-sign-Multipart_Sun_May__6_11:25:27_2007-1
Content-Type: application/pgp-signature
Content-Transfer-Encoding: 7bit

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
MessageID: h+tqmuwOXcJrFEjfW8iAS8GJ1eGeNKNB

iQA/AwUBRj3zaGZ9cbd4v/R/EQKWOQCg0UzZGC+hxH2GsttWJcH63Ym/EYIAn1nE
nHEaFteAkUp3lgzxyhHnAi6A
=3hFB
-----END PGP SIGNATURE-----

--pgp-sign-Multipart_Sun_May__6_11:25:27_2007-1--