Hard disk weirdness

[Rob Nicholson]

I've got a friend's laptop on my desk at the moment as it's behaving very
erratically. I recently rebuilt it for him (for another reason) but after
about two weeks flawless operation, it refused to boot. It then started
booting but was running vey slowly.

So I've got it back - I've re-formatted the hard disk using GDISK.EXE (part
of Norton's Ghost) and it didn't report any bad blocks.

However, I've left it running SCANDISK.EXE from Windows 98 and although no
bad clusters are being reported, it's taking an awfully long time to scan
the surface. Anyone got a rough idea how long it should take to SCANDISK
(surface scan) a 30GB notebook IDE drive? It's been going for about 8 hours
so far and is about 1/3 way through.

The other worrying thing is that the "examined clusters" counter isn't
counting up uniformly - it's surging, i.e. the stays on the same count for
about 2 seconds, then quickly counts up about 1,000 clusters, pauses for 2
seconds etc. Is this how SCANDISK works? I kind of expected that counter to
simple count up uniformly.

All a bit weird...

Cheers, Rob.

 

[Rod Speed]

I've got a friend's laptop on my desk at the moment as it's behaving very
erratically. I recently rebuilt it for him (for another reason) but after
about two weeks flawless operation, it refused to boot. It then started
booting but was running vey slowly.

So I've got it back - I've re-formatted the hard disk using GDISK.EXE (part of
Norton's Ghost) and it didn't report any bad blocks.

However, I've left it running SCANDISK.EXE from Windows 98 and although no bad
clusters are being reported, it's taking an awfully long time to scan the
surface. Anyone got a rough idea how long it should take to SCANDISK (surface
scan) a 30GB notebook IDE drive? It's been going for about 8 hours so far and
is about 1/3 way through.

Urk, thats completely hopeless.

The other worrying thing is that the "examined clusters" counter isn't
counting up uniformly - it's surging, i.e. the stays on the same count for
about 2 seconds, then quickly counts up about 1,000 clusters, pauses for 2
seconds etc. Is this how SCANDISK works?
Nope.

I kind of expected that counter to simple count up uniformly.

Yeah, thats what its sposed to do unless its restarting.

Most likely the drive is going bad and the pauses are due to
sectors that need to be retried to read them, but not bad enough
to be reported as bad, because they do read fine on retrys.

The other possibility is that the laptop itself is behaving weirdly
in that its deliberately slowing the cpu down when that makes no
sense. Modern laptops do that to minimise battery use when nothing
much is going on and its possible that mechanism is flakey. Can also
happen if it decides that the cpu temp is unacceptible too. Try running
SpeedFan and monitor the cpu temp and hard drive temp and see if
there is any obvious pattern there with pauses and the temps.

 

[Svend Olaf Mikkelsen]

I've got a friend's laptop on my desk at the moment as it's behaving very
erratically. I recently rebuilt it for him (for another reason) but after
about two weeks flawless operation, it refused to boot. It then started
booting but was running vey slowly.

So I've got it back - I've re-formatted the hard disk using GDISK.EXE (part
of Norton's Ghost) and it didn't report any bad blocks.

However, I've left it running SCANDISK.EXE from Windows 98 and although no
bad clusters are being reported, it's taking an awfully long time to scan
the surface. Anyone got a rough idea how long it should take to SCANDISK
(surface scan) a 30GB notebook IDE drive? It's been going for about 8 hours
so far and is about 1/3 way through.

The other worrying thing is that the "examined clusters" counter isn't
counting up uniformly - it's surging, i.e. the stays on the same count for
about 2 seconds, then quickly counts up about 1,000 clusters, pauses for 2
seconds etc. Is this how SCANDISK works? I kind of expected that counter to
simple count up uniformly.

All a bit weird...

Cheers, Rob.

To test a disk you will have to find a tool that runs from DOS, not
Windows or Linux.

 

[Rob Nicholson]

Urk, thats completely hopeless.

That's what I thought.

Most likely the drive is going bad and the pauses are due to

Shortly after posting, it did report two bad clusters/sectors and I selected
to fix them. However, scandisk hung at some point afterwards.

sectors that need to be retried to read them, but not bad enough
to be reported as bad, because they do read fine on retrys.

That makes sense - borderline failure.

happen if it decides that the cpu temp is unacceptible too. Try running
SpeedFan and monitor the cpu temp and hard drive temp and see if
there is any obvious pattern there with pauses and the temps.

The fan is temperature controlled as you can hear it coming on and off. I'm
guessing that the temperature being used to control this is the CPU
temperature. Assuming this is working okay, then I'd infer that the CPU
isn't running too hot.

Cheers, Rob.

 

[Rob Nicholson]

The other worrying thing is that the "examined clusters" counter isn't

Nope.

I've restarted scandisk this morning and it is indeed counting up uniformly
at the start of the disk. It's all pointing towards a failed/failing hard
disk.

Cheers, Rob.

 

[Rob Nicholson]

To test a disk you will have to find a tool that runs from DOS, not

I am - both gdisk and scandisk are DOS tools.

Cheers, Rob.

 

[Rob Nicholson]

I've restarted scandisk this morning and it is indeed counting up

uniformly at the start of the disk. It's all pointing towards a
failed/failing hard disk.

Nope, something is definately broken Running scandisk for the second
time and it counted uniformly for about 5 mins and then hung. Upon
rebooting, it's hung at 4 clusters.

I'm going to see if I can temporarily move a hard disk from one of the
company Dell laptops to make sure it is the hard disk and not something more
serious on the motherboard. Are drives typically compatible between laptops?

Cheers, Rob.

 

[Rod Speed]

That's what I thought.

Shortly after posting, it did report two bad clusters/sectors and I selected
to fix them. However, scandisk hung at some point afterwards.

Yeah, something is clearly dying in the hard drive subsystem,
normally the drive itself, but it can be the cable or the controller.

That makes sense - borderline failure.

And that's now confirmed by the report of a couple of bads.

The fan is temperature controlled as you can hear it coming on and off. I'm
guessing that the temperature being used to control this is the CPU
temperature. Assuming this is working okay, then I'd infer that the CPU isn't
running too hot.

I really just meant that that temperature sensing might
have failed for some reason. Not worth bothering with
that possibility now that scandisk has reported a couple
of bads tho, the drive is likely on its way out.

 

[Rod Speed]

Nope, something is definately broken Running scandisk for the second time
and it counted uniformly for about 5 mins and then hung. Upon rebooting, it's
hung at 4 clusters.

I'm going to see if I can temporarily move a hard disk from one of the company
Dell laptops to make sure it is the hard disk and not something more serious
on the motherboard. Are drives typically compatible between laptops?

Thats not necessarily a good idea because the OS will normally
attempt to update itself for the different hardware with a drive swap.

I'd personally test the drive in a desktop system as an extra drive
using a 3.5"/2.5" adapter, but then I have one of those so its easy.

 

[Rob Nicholson]

Thats not necessarily a good idea because the OS will normally

attempt to update itself for the different hardware with a drive swap.

I'd personally test the drive in a desktop system as an extra drive
using a 3.5"/2.5" adapter, but then I have one of those so its easy.

Unfortunately, I don't have access to one of those.

The fault is intermittent or maybe heat related. Yesterday, scandisk
wouldn't get beyond 4 clusters. This morning, after the laptop has been off
all night, it's started up again and is merrily checking the same sectors
that failed yesterday - it's also counting up uniformly. Currently at 10,000
clusters out of 1,220,640. If it did carry on at this rate, it would
probably scan the disk in an hour or so.

Still need to borrow a laptop hard drive.

Cheers, Rob.

 

[Rod Speed]

Unfortunately, I don't have access to one of those.

They're dirt cheap, so that may be one option.
http://www.compgeeks.com/details.asp?invtid=HD-108&cat=HDD

The fault is intermittent or maybe heat related. Yesterday, scandisk wouldn't
get beyond 4 clusters. This morning, after the laptop has been off all night,
it's started up again and is merrily checking the same sectors that failed
yesterday - it's also counting up uniformly. Currently at 10,000 clusters out
of 1,220,640. If it did carry on at this rate, it would probably scan the disk
in an hour or so.

Yeah, thats not at all uncommon with that sort of fault, that variable effect.

 

[Folkert Rienstra]

To test a disk you will have to find a tool that runs from DOS, not Windows or Linux.

Huh?

 

[Rob Nicholson]

Huh?

I think the OP meant that most low-level test tools appear to run from DOS.

Rob.

 

[Rob Nicholson]

Yeah, thats not at all uncommon with that sort of fault, that variable

effect.

Well the laptop has been working fine for over a week now. Nothing changed
really since it was failing all over the place last week.

Weird is the right title for this thread

I'm going to return it and see how the lad gets along. I've told him to "If
you install Limewire or Kazaa again, I'm not helping anymore" mainly as
these two products = virus infected PC.

Rob.

 

[Rod Speed]

Well the laptop has been working fine for over a week now. Nothing changed
really since it was failing all over the place last week.

Likely it just isnt getting as hot currently.

Weird is the right title for this thread

I'm going to return it and see how the lad gets along. I've told him to "If
you install Limewire or Kazaa again, I'm not helping anymore" mainly as these
two products = virus infected PC.

Its unlikely to be that, wouldnt produce the variable result.

 

[Rob Nicholson]

Its unlikely to be that, wouldnt produce the variable result.

I returned the laptop last week - will email him to see how he's getting
on...