How does a drive go bad when not in use?

[Rod Speed]

Yep.

That's actually one application where stiction may still
occur when the heads fail to be parked on the ramp.

Pathetic, really.

Nonsense. Plenty of drives still have no ramps.

Wrong, as always.

 

[helpdisc]

Ok so Your drive , the one u use now is
toshiba?
and the previus was ibm or Hitachi?
Or?

 

[bxf]

Ok so Your drive , the one u use now is
toshiba?
and the previus was ibm or Hitachi?
Or?

The old one (i.e. the original one in the laptop, the one being
discussed in this post) is a Seagate Momentus (4200RPM). The new one is
the 7200 RPM 7K60 from Hitachi/IBM.

 

[Folkert Rienstra]

The old one (i.e. the original one in the laptop, the one being
discussed in this post) is a Seagate Momentus (4200RPM).
The new one is the 7200 RPM 7K60 from Hitachi/IBM.

Oh please, stop that. Don't encourage that idiot.

 

[bxf]

Oh please, stop that. Don't encourage that idiot.

Hmmm, I must infer from your comment that I haven't been hanging around
here often enough to know who is who and what is what. I DID wonder
about the possible usefulness of the question/answer.

 

[Folkert Rienstra]

Hmmm, I must infer from your comment that I haven't been hanging
around here often enough to know who is who and what is what.

You don't need to. He came in here about the same time you did.
(And oh, nothing implied with that of course ;-))

 

[bxf]

Well, this drive is obviously ****ed. No big deal, I suppose, except
for the fact that I don't understand how it breaks while not in use.

I divided it (40BG) into 4 partitions and installed Windows, just so
that I can put the drive directly into the laptop, rather than using it
via USB. Two of the partitions are OK, and the other two won't let me
format or do error checking with recovery of bad sectors.

For what it's worth, here is some SMART data:


Threshold Value Worst Data
--------- ----- ----- -----
Raw READ error rate 34 54 54 47863983
Spin up time 0 99 98 0
Start/stop count 20 98 98 2794
Reallocated sector count 36 100 100 1
Seek error rate 30 74 60 28704318
Power on time count 0 98 98 1994
Spin retry count 34 100 100 0
Power cycle count 20 100 100 719
Power-off retract count 0 100 100 145
Load/unload cycle count 0 78 78 45140
Temperature 0 46 57 46
Hardware ECC recovered 0 54 53 47863983
Current pending sector count 0 100 100 71
Off-line uncorectable sector count 0 100 100 71
Ultra ATA CRC error rate 0 200 195 12
Write error rate 0 100 253 0
<Vendor specific> 0 100 253 0

After a bit of use, the 47863983 figures went up to over 53 million (in
both places), and the "Seek error rate" went up to 28742257.

 

[Rod Speed]

Well, this drive is obviously ****ed. No big deal, I suppose, except
for the fact that I don't understand how it breaks while not in use.

Some faults are use dependant, some arent. Dry
joints and cracked traces etc arent use dependant.

I divided it (40BG) into 4 partitions and installed Windows, just
so that I can put the drive directly into the laptop, rather than using
it via USB. Two of the partitions are OK, and the other two won't
let me format or do error checking with recovery of bad sectors.

For what it's worth, here is some SMART data:

Threshold Value Worst Data
--------- ----- ----- -----
Raw READ error rate 34 54 54 47863983
Spin up time 0 99 98 0
Start/stop count 20 98 98 2794
Reallocated sector count 36 100 100 1
Seek error rate 30 74 60 28704318

That may or may not indicate a dying drive, depending on the drive.

Power on time count 0 98 98 1994
Spin retry count 34 100 100 0
Power cycle count 20 100 100 719
Power-off retract count 0 100 100 145
Load/unload cycle count 0 78 78 45140
Temperature 0 46 57 46
Hardware ECC recovered 0 54 53 47863983

Current pending sector count 0 100 100 71
Off-line uncorectable sector count 0 100 100 71

Those two certainly indicated a dying drive.

Ultra ATA CRC error rate 0 200 195 12
Write error rate 0 100 253 0
<Vendor specific> 0 100 253 0

After a bit of use, the 47863983 figures went up to over 53 million
(in both places), and the "Seek error rate" went up to 28742257.

She's dead Jim, you into necrophilia ?

 

[bxf]

Some faults are use dependant, some arent. Dry
joints and cracked traces etc arent use dependant.




That may or may not indicate a dying drive, depending on the drive.



Those two certainly indicated a dying drive.



She's dead Jim, you into necrophilia ?

Never tried it.

Actually, I've got two partitions of about 10GB each that appear to be
useable without problems (for now). And, if I was really determined on
this matter (I'm not), I could expand the good partitions until I hit a
bad area, thus maximizing useable space.

I do have more interesting things to do, though.

 

[Folkert Rienstra]

Never tried it.

Actually, I've got two partitions of about 10GB each that appear to be
useable without problems (for now).

That's assuming the drive doesn't have more structural problems ...

And, if I was really determined on
this matter (I'm not), I could expand the good partitions until I hit a
bad area, thus maximizing useable space.

..... so then why not let the drive itself take care of the bad areas.

 

[Rod Speed]

Rod Speed wrote

Never tried it.

A likely story.

Actually, I've got two partitions of about 10GB each that appear
to be useable without problems (for now). And, if I was really
determined on this matter (I'm not), I could expand the good
partitions until I hit a bad area, thus maximizing useable space.

That assumes that the bad bits have a fixed location
on the platter. Thats unlikely with most faults, particularly
with faults that develop with the drive not being used.

I do have more interesting things to do, though.

Yeah, very unlikely to be a useful use of your
time given how cheap hard drives are now.

 

[bxf]

A likely story.

I swear it's true. However, I can attest to the existence of some women
who needed close examination to ascertain the fact that they were not
in fact dead.

That assumes that the bad bits have a fixed location
on the platter. Thats unlikely with most faults, particularly
with faults that develop with the drive not being used.

Just out of interest, can you explain this one? I have difficulty
visualizing a bad disk with a "floating" bad area (assuming the problem
is not with the heads).

 

[Rod Speed]

Rod Speed wrote

I swear it's true.

Corse you would say that.

However, I can attest to the existence of some women who needed
close examination to ascertain the fact that they were not in fact dead.

Yeah, can be a problem.

Just out of interest, can you explain this one? I have
difficulty visualizing a bad disk with a "floating" bad
area (assuming the problem is not with the heads).

Clearly any fault that isnt due to specific bits of the
platter going bad will see the bad sector reported
essentially randomly placed on the platters and the
report on which sectors are bad varying randomly over time.

Even with the heads, there is no reason why
most faults should result in the same sectors
being reported as bad over time.

 

[bxf]

Clearly any fault that isnt due to specific bits of the
platter going bad will see the bad sector reported
essentially randomly placed on the platters and the
report on which sectors are bad varying randomly over time.

OK, I just viewed a bad drive as either a platter problem or a head
problem, or both. So of course, a platter problem is in a fixed
location (unless the heads are rubbing against the platter and
propagating the problem, but that's another story).

Even with the heads, there is no reason why
most faults should result in the same sectors
being reported as bad over time.

This one is pretty obvious, I'd think. If the heads are bad then the
errors are likely to occur at no specific place.

 

[Rod Speed]

Rod Speed wrote

OK, I just viewed a bad drive as either a
platter problem or a head problem, or both.

That is just plain wrong, most obviously with logic card faults.

So of course, a platter problem is in a fixed location
(unless the heads are rubbing against the platter and
propagating the problem, but that's another story).

It isnt really, because about the only way a platter
can go bad in a particular area of the platter is if
you get a head crash at that spot, or the platter
starts shedding the magnetic surface. Both of those
failures will normally produce effects outside the
area that was initially the problem, with loose
material floating around inside the sealed
enclosure, or getting stuck on the heads etc.

This one is pretty obvious, I'd think. If the heads are bad
then the errors are likely to occur at no specific place.

That isnt always true. If the gain starts to deteriorate,
the poorest performing areas of the platter magnetically
would normally show up as bads first. You can get the
same effect if its stops flying properly too so it no
longer flys as close to the surface reliably.

And a fault in the flexible connection to the heads can see the
fault more of a problem in some head positions than others too.

The short story is that you dont normally see particular parts
of the platter just go bad and you can just spare those sectors
and carry on regardless with no risk of more bads showing up.
The basic physics means that that just doesnt happen.

The automatic sparing mechanism that modern drives have is
there for a different reason, marginal areas of the platters that
look usable in the manufacturing test turning out to be not really
reliably usable, so they just get marked bad in the field when
ideally they should have been marked bad in the factory initially.

 

[larryking]

www.pathtoodeep.com fixes that problem