PCBUILD Archives

Personal Computer Hardware discussion List

PCBUILD@LISTSERV.ICORS.ORG

Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Tom Turak <[log in to unmask]>
Reply To:
PCBUILD - Personal Computer Hardware discussion List <[log in to unmask]>
Date:
Tue, 3 Feb 2004 12:07:53 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (72 lines)
Peter,
While everything you say is true, I guess Michele has had the same trouble I
have come across on several occasions.  Disk testing and repairing software
is only as reliable as the system it is running on.  The leading cause of
problems with drives in my own experience has been with corrupt memory,
either due to a bad chip, or to a bad program (driver or otherwise, I don't
see this anymore now that I use Win2000 / XP, the last two times it was a
bad memory chip module).  Testing a drive involves copying the contents to
memory, then re-reading the drive and making comparisons to memory.
Statistics are captured as to how many re-reads were identical. You can
quickly see that if the memory is malfunctioning, the tests can be
compromised and successful re-reads will be graded as failures because the
memory contents changed, not the disk contents (you would suspect a failing
disk to randomly misread the data stored on it, you don't normally suspect
the memory is doing the same thing).

My worst cases have been hardware drivers or faulty chips that damage the
file system, not the physical disk.  My first response is to test the drive
to see if I can find out why the file system is being damaged (corrupt word
files, garbled folder listings, missing files, windows crashes).  Tests will
report the drive has bad clusters.  But when I move the drive to a known
working pc, there the drive tests out okay.

When the memory problem is corrected, the disk errors go away and the pc
stability returns.  Now when you run a scan, Ta da! there are no bad
sectors.  The real horror to this story is if you don't fix the memory, but
continue to run scans, more and more of the disk clusters become tagged as
bad and are removed from the file system.  Then you have to run a serious
scan to test and return them to the usable disk capacity.

Of course, what you said could be the right explanation.  Physical disk
damage is not unheard of, so error messages reporting bad sectors can be the
first signs of a failing disk. I'm just pointing out what has happened more
often than physial damage in my own experience, especially when I had to
deal with win9x software.
Tom Turak

-----Original Message-----
From: [log in to unmask] [mailto:[log in to unmask]]
Sent: Tuesday, February 03, 2004 7:26 AM

>In a message dated 2/3/2004 2:31:46 AM Eastern Standard Time,
>[log in to unmask] writes:
>I have to agree on this one, but only because of personal experience.
>I have a drive in a system which was reporting bad sectors........
>I ended up formatting/reloading the o/s and running the scan again
>continuously just to see what would happen.
>Strangely enough, the bad sectors were gone, and haven't reappeared to this
>day.
>I'm assuming that the scans don't actually know the difference between
>'physical' damage to the drive and 'software' damage. This seems the only
>explaination........
>Michele Sayer

Hi,
  Drive makers include "spare" sectors, to replace sectors that go bad. That
is why the bad sectors "disappeared" when you did the thorough scandisk. It
found the bad sectors and marked them as such, and the disk replaced them
with
spare sectors, so you didn't lose any disk capacity. (This works until you
exhaust the supply of "spare" sectors, but if you have that many bad
sectors, the
drive is probably soon to be toast anyway.)

HTH,
Peter Hogan
[log in to unmask]

                         PCBUILD's List Owners:
                      Bob Wright<[log in to unmask]>
                       Drew Dunn<[log in to unmask]>

ATOM RSS1 RSS2