Subject: Re: my HD probs...
To: Michael G. Schabert <mikeride@prez.org>
From: Miles Nordin <carton@Ivy.NET>
List: port-alpha
Date: 04/13/1999 19:50:47
yeah the Sad Mac.  I sound like a broken record, but: 

 1. Check the SCSI ID.
 2. Check the Termination.
 3. Use the Startup Disk control panel, make sure your good disk
    is the first one the ROM tries.
 4. Check the Termination again, more carefully.  In my experience,
    nobody actually checks the termination the first time you ask them to.

if you're trying to resurrect a failing hard drive, the best tool i've
found for this (on any platform) is FWB SCSI Configure.  _not_ the ``Hard
Disk Toolkit'' icon--Configure is a seperate application.  It lets you
tweak and query SCSI2 mode pages.  You can, for example, get an idea of
how much the defect list has grown since the drive was manufactured, turn
on various auto-recovery, auto-remapping features, and control how many
soft errors the drive will tolerate before passing an error up to the OS.
You can tweak the mode pages to be a little less conservative, and then
use regular Toolkit to low-level the drive.  I believe FWB includes a fair
amount of AppleGuide docs or something about what the different pages
mean, if it isn't intuitive.

``Verification'' in Toolkit is, as i understand it, a load of BS.  At best
it's a nice thing to do last, after you've got a drive that you think
works.  But if you tell a drive to lowlevel itself, so-called
``verification'' is necessarily part of the process; the only
justification for doing it twice is supposed firmware bugs in the drive.
And this is rather presumptuous, in my opinion, especially for a MacOS
app. Besides, verification really needs to happen _inside_ the drive since
a good verification has a certain sort of analog character to it.

In my experience, hard disks are brilliant little pieces of Singaporean
junk.  All the confusion comes from OS's like Linux that go into
``resetting SCSI bus...'' busywait-laden loops when drives dutifully
report perfectly sensical errors.  The philosophy that any hard disk which
reports any error must be completely insane and should be dealt with
ruthlessly with lots of electroshock therapy and brutally repetitive retry
cycles, is just plain wrong.  And, to be honest, this applies to your
effort to ``rescue'' the drive as well.

If you're really desperate to get a few more weeks of use out of the drive
before it crashes again and loses all your data, you'll probably go
through several configuretweak-lowlevel passes before you give up, and
there's no point in verifying every time.

Suggestion:  if the thing is less than five years old, send it back to
Quantum for a free replacement.  I don't know how to do this, but others
have done it, and it's routine.

-- 
Miles Nordin / 1-888-857-2723
555 Bryant Street #182 / Palo Alto, CA 94301-1700