More life from hard disk with bad sectors

[Ardent]

Hi All

One of my hard disks continued to develop bad sectors and every time I
was using format to mark the bad sectors. Suddenly I got an idea of
not only marking the bad sector but also keeping the head from
accessing areas around the bad sector.

By the time I did this to my hard disks it had only 1 gb of its 2 gb
capacity usable. (You can imagine the number of bad sectors!).

The bad sectors developed during a three month period. After marking
the areas so the read head did not go anywhere near the bad sectors I
have been using this hard disk for more than two years without any new
bad sector appearing!

You can read my experience here

http://members.tripod.com/~diligent/ninelives.htm

HTH

 

[Michael-NC]

For Christ's sake, just throw the dam thing away. Who in their right mind
wants a 2GB hard drive that's rife with bad sectors???

Please, go away and don't come back!

 

[Ron Reaugh]

Hi All

One of my hard disks continued to develop bad sectors and every time I

As soon as a HD begins to develop more than a very few bad sectors then
assume the HD is DOA as the rest will go bad soon. Replace the HD. We'll
keep your post as perfect proof of this point.

 

[Jim Berwick]

As soon as a HD begins to develop more than a very few bad sectors
then assume the HD is DOA as the rest will go bad soon. Replace the
HD. We'll keep your post as perfect proof of this point.

I consider anything > 0 bad sectors as meaning "The drive is dying,
immediately replace it"

 

[Eric Gisin]

Nonsense. Your drive had bad sectors when it was shipped.

 

[Monster]

lol, back in the old days (286, 386) harddrives weren't cheap so when an hd
developed bad sectors, the bad sector areas was partitioned off and never
used again

 

[CJT]

Nonsense. Your drive had bad sectors when it was shipped.

Not that are visible. By the time you see any, there are really
quite a few. That's different from a few random ones at the time
of manufacture.

 

[J. Clarke]

lol, back in the old days (286, 386) harddrives weren't cheap so when an
hd developed bad sectors, the bad sector areas was partitioned off and
never used again

Still are. But it's handled transparently by drive now. When an IDE drive
is showing bad sectors it generally means that it's got more than the
onboard sparing can handle, which means that it's _real_ bad.

 

[Eric Gisin]

Not that are visible. By the time you see any, there are really
quite a few. That's different from a few random ones at the time
of manufacture.

There are hundreds of initial bad sectors on each of my three SCSI drives.

It is quite normal to develop a few new ones each year, and that does NOT
indicate the drive is failing. Normally they are not visible, because they are
corrected and remapped.

 

[Al Dykes]

There are hundreds of initial bad sectors on each of my three SCSI drives.

It is quite normal to develop a few new ones each year, and that does NOT
indicate the drive is failing. Normally they are not visible, because they are
corrected and remapped.

SMART monitering will tell you when if the device is within
manufacturer's tolerances. Everest from Lavalist is a nice package
(free for noncommercial use.) I don't have a link handy.

With XP I've seen an event log entry following a perceptable pause on
my PC. The log showed that a block had just meen remapped.

If I see a single additional block remapped on a machine being used
for revenue-producing business use I plan to swap that disk
out, asap.

I'd mark the disk and use it for some don't-care-if-burns purpose.
The disk doesn't get a third chance.

 

[Eric Gisin]

Still are. But it's handled transparently by drive now. When an IDE drive
is showing bad sectors it generally means that it's got more than the
onboard sparing can handle, which means that it's _real_ bad.

There are around 100,000 spares on a 50GB drive. SMART would have said the
drive was toast long before they ran out.

 

[Folkert Rienstra]

Hi All

One of my hard disks continued to develop bad sectors and every time I
was using format to mark the bad sectors. Suddenly I got an idea of
not only marking the bad sector but also keeping the head from
accessing areas around the bad sector.

By the time I did this to my hard disks it had only 1 gb of its 2 gb
capacity usable. (You can imagine the number of bad sectors!).

The bad sectors developed during a three month period. After marking
the areas so the read head did not go anywhere near the bad sectors I
have been using this hard disk for more than two years without any new
bad sector appearing!

You can read my experience here

http://members.tripod.com/~diligent/ninelives.htm

Several possible flaws in your method.

1) only works for drives smaller than 8GB

2) no drive has 256 physical heads and 63 sectors per track so
1 logical cylinder is already many physical cylinders
With 4 heads and ~600 sectors/tr. you already reserve ~6
physical cylinders per logical cylinder.

3) it's not the heads that need keeping away from the damaged
spots but the slider that they are mounted on, which is much
wider. The slider can cover into the hundreds of cylinders.

Btw, why not finish the experiment and zero-write the whole
drive and use it as one normally would, and see what happens.

If your theory is correct it should soon die. Somehow, I doubt that.

 

[Folkert Rienstra]

Still are.
Nope.

But it's handled transparently by drive now.

But doesn't reduce its capacity.

When an IDE drive
is showing bad sectors it generally means that it's got more than the
the onboard sparing can handle, which means that it's _real_ bad.

Utterly Clueless.
When an IDE drive is showing bad sectors it generally means that
the "bad" sectors are unrecoverable read error bad sectors and
that the sparing system is incapable of replacing them as bad data
then would reappear as correct data in the replacement sector.

The unrecoverable read error bad sector will disappear nicely after
a write to the so-called "bad" sector.
"When an IDE drive is showing bad sectors it generally means" that
such a write to the 'so-called' "bad" sectors has not yet happened.

Ofcourse "J. Clarke" knows this as it has been explained to him
before but he prefers to troll.

 

[Folkert Rienstra]

I consider anything > 0 bad sectors as meaning "The drive is dying,
immediately replace it"

Well, maybe you should look into your S.M.A.R.T. data then, every hour.
It might be in it's death throws already and you not even know it.

 

[Roger Blake]

Nonsense. Your drive had bad sectors when it was shipped.

Modern hard drives do bad block forwarding, invisible to the operating
system and thus the end user. If you start actually seeing bad blocks
it means that all the spares have been used and the drive is in the
process of dying a grievous death.

 

[Taed Wynnell]

Modern hard drives do bad block forwarding, invisible to the operating
system and thus the end user. If you start actually seeing bad blocks
it means that all the spares have been used and the drive is in the
process of dying a grievous death.

I don't think that's quite right based on my understanding. If you're
WRITING to the block when the error is first detected by the drive, then the
reallocation is invisible to the OS. However, if a READ is the first time
the error is detected, then an error is returned to the OS (and it will
"detect" a bad block), but the disk then places that block number in a
pending list and the reallocation isn't done the next time it is written to.
Thus, the OS could see an error before the spares have been exhausted.

This process is described in most detailed data sheets for the drives, and
the 5 or so that I've read seem to follow this general process.

 

[Eric Gisin]

Clueless gibberish. Read my and Folkert's reply to J Clarke.

 

[CJT]

I don't think that's quite right based on my understanding. If you're
WRITING to the block when the error is first detected by the drive, then the
reallocation is invisible to the OS.

Are you implying drives do a read-after-write check? If not (and I
don't think they do such a check routinely), it seems to me the only
way a drive would ever detect such an error (i.e. upon a write) is
if things are so hosed the drive can't even sync to the write location.

However, if a READ is the first time

 

[Wizard]

My experience is that Blake is correct. Bad blocks, once seen, grow like
weeds.

 

[J. Clarke]

I don't think that's quite right based on my understanding. If you're
WRITING to the block when the error is first detected by the drive, then
the
reallocation is invisible to the OS. However, if a READ is the first time
the error is detected, then an error is returned to the OS (and it will
"detect" a bad block), but the disk then places that block number in a
pending list and the reallocation isn't done the next time it is written
to. Thus, the OS could see an error before the spares have been exhausted.

This process is described in most detailed data sheets for the drives, and
the 5 or so that I've read seem to follow this general process.

If one is being pedantically correct then yes, it's possible to see bad
sectors before the sparing is exhausted. In the real world if you're
seeing bad sectors on an IDE drive, assume the drive is in the transition
phase between "merely dead" and "sincerely dead".