Date: Mon, 27 Nov 1995 11:30:15 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: msmith@atrad.adelaide.edu.au (Michael Smith) Cc: terry@lambert.org, msmith@atrad.adelaide.edu.au, hasty@rah.star-gate.com, hackers@FreeBSD.org Subject: Re: Help! I got a bad block.... Message-ID: <199511271830.LAA19200@phaeton.artisoft.com> In-Reply-To: <199511270122.BAA02417@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Nov 27, 95 01:22:46 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > I don't buy the "on a bad day argument". I can't see how a good sector > > could be incorrectly marked bad, unless you have the driver cooperate in > > the marking and blow the driver programming. > > If the drive has forwarding enabled, and for one reason or another (thermal, > LF vibrations, wrong phase of the moon) has problems writing a sector, it'll > forward it, and you'll lose the original. Uh, how do you propose to force the moon phase to change do you can recover the original??? > Forwarding doesn't do all the job, as far as media perfection is concerned - > once you run out of spare sectors, you're screwed. I've met a few drives > that appear to have lost their marbles when they ran out, and others that > just start to report errors, despite the fact that forwarding is enabled. They are telling you "replace me, replace me". Typically this is a rephrase of the subliminal message "never buy this type of drive again, never buy this type of drive again". > Unless you can positively identify the drive, determine the number of _free_ > spare sectors, and make a judgement call on how many you're likely to use > in a given uptime period, you still need a higher layer to catch the failure > case. I think that having an adaptive layer that's SCSI specific for the three important cases (just turn it on, turn it on and cooperate, and don't turn it on, use commands to let the driver force the forwarding) is the way to go. I know that if I had a striping setup, I would *never* want the drive to remap the sectors. On the other hand, the default behaviour you can install doesn't include striping and doesn't include cooperative remapping by the SCSI driver layer. So turning it on is a win until such time as the changes the in kernel code occur. Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199511271830.LAA19200>