From owner-freebsd-questions Wed Jul 4 18:55:21 2001 Delivered-To: freebsd-questions@freebsd.org Received: from dan.emsphone.com (dan.emsphone.com [199.67.51.101]) by hub.freebsd.org (Postfix) with ESMTP id 9439237B405 for ; Wed, 4 Jul 2001 18:55:18 -0700 (PDT) (envelope-from dan@dan.emsphone.com) Received: (from dan@localhost) by dan.emsphone.com (8.11.3/8.11.3) id f651tF214281; Wed, 4 Jul 2001 20:55:15 -0500 (CDT) (envelope-from dan) Date: Wed, 4 Jul 2001 20:55:15 -0500 From: Dan Nelson To: Dale Hagglund Cc: freebsd-questions@FreeBSD.ORG, wmoran@iowna.com Subject: Re: SCSI bad block remapping Message-ID: <20010704205514.A13653@dan.emsphone.com> References: <3B327713.46173EC5@iowna.com> <863d8curun.fsf@ponoka.battleriver.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <863d8curun.fsf@ponoka.battleriver.com> User-Agent: Mutt/1.3.19i X-OS: FreeBSD 5.0-CURRENT Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG In the last episode (Jul 04), Dale Hagglund said: > I work with some guys who've done weird things with and to scsi > drives for quite a few years, and, according to them, turning on > automatic sector remapping can have surprising and usually > undesirable results. > > The two key issues are > > (a) the disk might do remapping in fairly large chunks, even > up to a track in size, and This is easy to verify; just run "camcontrol da0 defects -f phys -G" and see if entire cylinders get remapped. The couple drives I've looked at map single blocks. Contaminants in the drive case would tend to produce damage that takes out entire tracks, so this might be what your friends saw. > (b) after remapping, the data in the remapped region is > undefined. I am extremely skeptical of this. If the data can be remapped, it is remapped correctly (that's what all the ECC stuff is for). If it cannot be remapped, an error is returned. Failure to do this should be easily detectable, and if any vendor shipped drives that clamed to do AWRE/ARRE and didn't we'd see it on the news somewhere. RAID units disable A?RE not because it's unreliable but because of the performance hit and because in a RAID it's better to remove a drive at the first sign of failure than later. All of the drives in my home systems are "failed" disks kicked out by RAID arrays at work that I simply reformat and flip A?RE on. -- Dan Nelson dnelson@emsphone.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message