[1734] in linux-scsi channel archive
Re: enabling error recovery
daemon@ATHENA.MIT.EDU (Michael Weller)
Mon Apr 21 05:51:34 1997
Date: Mon, 21 Apr 1997 11:47:12 +0200 (MESZ)
From: Michael Weller <eowmob@exp-math.uni-essen.de>
To: Jim Van Zandt <jrv@vanzandt.mv.com>
Cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <199704210044.UAA00273@vanzandt.mv.com>
On Sun, 20 Apr 1997, Jim Van Zandt wrote:
> I calculate that there are 8205*6=49230 tracks in all, so there must
> be only one zone.
Yep.
> How much of the disk should I set aside? My other scsi disk has one
> alternate sector for every 765. That would suggest about 64 alternate
> tracks for this disk, or 64*171=10944 alternate sectors.
I've no good hint here. But 5 MB alternate sectors sounds like a sensible
size to me.
> Should I set "alternate tracks per zone", or "alternate sectors per
> zone", or both? What are the advantages and limitations of alternate
> tracks vs. alternate sectors?
Hmm, I'd suspect that alternate sectors is a bit better because the
replacement sector might be close to a faulty one in case of a remap since
the alternate sectors are probably spread wider over the disk than
alternate tracks would be. In case of no remaps, the disks is probably
faster with alternate tracks because there are no short tracks on the disk
then (less head moves).
I won't be surprised if this question is only theoretical. Probably the
disk will either only support alternate sectors or tracks. Try which
setting it accepts and use that.
> Suppose I set "alternate tracks per zone" to 64. Then I have my scsi
> host adapter (a BusLogic 948) do a low level format, right?
Yes, but you also need to save this to the NVRAM of the disk with
scsi-info b4 low level formatting.
> This disk already has a lot of defects:
> Data from Defect Lists
> ----------------------
> 889 entries in manufacturer table.
> Cyl Head Byte_offset
> 54 4 52224
> 54 4 52736
> ...
>
> I assume that those sectors have already been reassigned.
Yep. And actually already at manufacturing time.
> If I do a low-level format, will they still be reassigned?
Basically, this depends on the low-level SCSI-format command which that
low-level format utility uses. It can ignore a list, just keep, or add new
defects during the format (either given manually or newly found ones). It
can even make extensive checks to add all not 100% perfect spots on the
disk to the defect list (media certify).
I don't think there is a way to ignore the manufacturers list though. I'd
also suspect your low level formatter will be sensible and not erase the
grown defects (if any). In case you have the time, want it 100% and
distrust the disk somehow, try making a disk certify with your low level
formatter (most I saw support that).
Michael.
(eowmob@exp-math.uni-essen.de or eowmob@pollux.exp-math.uni-essen.de
Please do not use my vm or de0hrz1a accounts anymore. In case of real
problems reaching me try mat42b@spi.power.uni-essen.de instead.)
