>I've managed hundreds of disk problems, of which only a handful were 
>hardware problems:

I've seen lots of hardware problems, BUT, it should be noted that I also 
have worked with potentially thousands of drives, many of which are in 
excess of 10 years old and have been well abused.

When I see hardware problems, it is usually one of 3 things (in order 
that I see them): 1. Drive fails to spin up from a dead (or dying) motor. 
That is the single most common hardware failure I see. Next would be 2. 
Head "stiction". That is when the heads either adhere to the drive 
platters due to years of grime built up inside (even assembled in a clean 
room, gunk will still make it into the drive and it WILL build up on the 
heads as they spin). When the drive is off and cools down, this normally 
harmless gunk becomes a mild glue, and literally glues the heads to the 
platter. Sometimes this can be cured by tapping the drive to unstick it, 
other times its a loss. I also lump into this category failed head 
stepper motors. Although technically it is a different fault, I count it 
with the stuck heads because in many cases stiction is what causes the 
stepper motor to burn out.

And finally 3: Controller failure. The smarts on the drive die. Usually a 
fried chip or other part. IF you have an identical drive, you can usually 
swap out the controller board to revive the drive. However, you have to 
know that it is really the exact same drive. Its not uncommon for a 
company to go thru different revisions of a drive during a drive's line 
life. It is not always safe to swap the controller from a different 
revision.


With all the drives I've dealt with, I've only ever seen a handful of 
truly crashed heads (where the head either hits the platter, and/or is 
ripped off the arm). The few times I've seen it, you usually know it as 
it makes a pretty loud grinding noise (as would be expected, since you 
are now dragging metal on metal at a few thousand RPM).


And beyond all the hardware failures I have seen, I have seen far more 
media failures. Technically they can be lumped into hardware as well, 
since the drive platters are technically hardware. However, since mild 
media failures can be corrected with software, I prefer to consider them 
software failures. As a drive ages, the magnetic media on the platters 
starts to fail. Its just physics. Magnetism will NOT last forever. As it 
decays, the drive becomes more and more unreliable. Eventually enough bad 
sectors crop up that the drive is no good. Also, finding bad sectors in a 
drive out of the blue is usually a sign that the media is starting to 
fail. Sometimes its a random occurrence and the drive will continue to 
operate fine for years and years to come... other times it is just the 
beginning of a cascade failure, and more sectors will die out very 
shortly.

And of course, the #1 thing I have found wrong with drives is just simply 
corrupted data. These are 100% software related. Usually due to constant 
system crashing, and the computer writing out incorrect data at the wrong 
place and wrong time. These can be anywhere from the relatively harmless 
things like bad file dates, thru slightly worse things like bad drive 
tables, thru the really bad things like totally corrupted formats. These 
are all almost always 100% fixable in terms of reusing the drive. A new 
format will almost always correct the problem (I hesitate to say will 
always correct it, as as soon as I do, someone will come up with an 
instance that says I'm wrong... but I've personally never lost a drive to 
this kind of failure). However, even though the drive can be saved, it 
doesn't always mean you can save the data off the drive. Sometimes it is 
lost beyond what any commercially available tools can get at. But a 
software failure is almost always trivial for a professional recovery 
shop to salvage. (and it getting better and better at the home user being 
able to do it as well as better and better tools are being sold to do it).


Not sure anyone cared, but since I'm trying to avoid doing a long, 
boring, complicated job here at work, I figured I'd waste some time and 
chime in with my experiences.

-chris
<http://www.mythtech.net>

___________________________________________________________________________
To unsubscribe send a mail message with a SUBJECT line of "unsubscribe" to
<[EMAIL PROTECTED]>  or  <[EMAIL PROTECTED]>

Reply via email to