Hard Disk requalification

[Ian East]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Thanks for any help.

 

[William W. Plummer]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

What does "re-qualification" mean? I believe you can run a program to
say how many sectors are good or bad. And you can verify that there
are no errors during an X-hour run. But when I purchase a drive, I'm
looking for an indication that the drive will continue to serve me for
some number of years in the future. I want to know how many hours are
on the bearings, whether it has been operated in a high temperature
environment that might have left the electronic components dried out.

 

[kony]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Thanks for any help.

What makes you feel they would need replaced with *same*
thing, if they were to fail (since you wrote "not easily
replaced") ?

If they are pretty old it doesn't necessarily matter if they
pass a test, their expected lifespan is still up. Suppose
you had a set of bald tires... just because the car stops,
because you can cruise at highway speeds, does that mean
it's a good idea to keep using them until you find they
CAN'T cruise at highway speeds?

The HDD manufacturer should have diagnostics that will check
a drive.

Here are some,
http://www.tacktech.com/display.cfm?ttid=287

 

[Bob Day]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Thanks for any help.

Download diag378.exe from ftp://ftp1.dell.com/diags/ and see
whether it fits your needs.

-- Bob Day
http://bobday.vze.com

 

[Ian East]

What makes you feel they would need replaced with *same*
thing, if they were to fail (since you wrote "not easily
replaced") ?

They have a proprietary OS image that is dependent on the physical
geometry of the disk. Purchasing "new" refurbs from e-bay and whatnot
is not much different than using the disks we have now.

If they are pretty old it doesn't necessarily matter if they
pass a test, their expected lifespan is still up. Suppose
you had a set of bald tires... just because the car stops,
because you can cruise at highway speeds, does that mean
it's a good idea to keep using them until you find they
CAN'T cruise at highway speeds?

Agreed... But but creating a test that would cause an existing
physical disk problem to manifest itself now rather than later is the
idea. If you find that replacing the bald tires on your car is going
to cost you far more than the car is worth, you might want to wait a
little longer until next years budget and you can afford to buy a new
car.

The HDD manufacturer should have diagnostics that will check
a drive.

I have tried that and the test does not actually write to the disk
then read. Every sector must be written to in order to truly know if
the disk is failing.

 

[Ian East]

What does "re-qualification" mean? I believe you can run a program to
say how many sectors are good or bad. And you can verify that there
are no errors during an X-hour run. But when I purchase a drive, I'm
looking for an indication that the drive will continue to serve me for
some number of years in the future. I want to know how many hours are
on the bearings, whether it has been operated in a high temperature
environment that might have left the electronic components dried out.

Re-qualification means putting the disk through approximately the same
tests that the manufacturer uses to qualify the disk in the first
place. Seeing how many sectors are good or bad is not quite what
we're after. What we're looking for is some way to put a disk through
a run, get some statistical data, then try to determine the
probability that a disk is going to fail. Disk age is a useful
statistic, but in this case we have not been able to determine a
strict relationship between age and disks that have failed in the
past.

 

[Ken]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Thanks for any help.

You might look into Spinrite.

 

[kony]

They have a proprietary OS image that is dependent on the physical
geometry of the disk. Purchasing "new" refurbs from e-bay and whatnot
is not much different than using the disks we have now.

"Dependent"?
What exactly is this OS image?

Agreed... But but creating a test that would cause an existing
physical disk problem to manifest itself now rather than later is the
idea.

Fair enough, but given you already feel this potential
exists, is it prudent to deploy systems with these old
drives?

Seems like a liability to me, either of data or a lawsuit or
??? Depends on the use.

If you find that replacing the bald tires on your car is going
to cost you far more than the car is worth, you might want to wait a
little longer until next years budget and you can afford to buy a new
car.

No, that's reckless and risks your life, maybe others' as
well. Wouldn't someone poor enough (or with different
priorities) that they have a car worth less than a set of
tires be too poor/whatever to be buying a new car at the
beginning of the next year? If there is budget for a new
car, tires costing little more than one, "maybe" two month's
payments are not such a burden. Come to think of it the
insurance alone on the new car is going to cost enough that
new tires are still a better solution if they're only used
for a month or two, which would help to sell the car as
well, or rather, might make it sellable at all. I've gone
off on a tangent but I still disagree completely with your
concept of testing something old enough that it's inevitable
that it fail sooner rather than later.

I have tried that and the test does not actually write to the disk
then read. Every sector must be written to in order to truly know if
the disk is failing.

Then how do you expect to keep that proprietary OS image?
Write testing generally is done to a blank (or soon-to-be)
drive.

If this OS image is valuable I'd think you should be keeping
an archival backup of it, and it would be even more prudent
given the propensity for failure even while doing the
testing. If you make such a backup with Driveimage or Ghost
you should then have either of them read and make a backup
image (while there is still greatest chance for intact data)
then restore that image using the write-verify option the
programs offer.

 

[John McGaw]

You might look into Spinrite.

http://www.grc.com/sr/spinrite.htm

 

[Shep©]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Thanks for any help.

As per other post,"Spinrite" or the free,"Powermax",
http://tinyurl.com/4osje
Although made by Maxtor will run on most makes(I've not found one yet
it won't do).
HTH




--
Free Windows/PC help,
http://www.geocities.com/sheppola/trouble.html
remove obvious to reply
email (e-mail address removed)
Free original songs to download and,"BURN" :O)
http://www.soundclick.com/bands/8/nomessiahsmusic.htm

 

[VWWall]

Hi,

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

Try SpinRite, Steve Gibson's program.

http://www.grc.com/sr/spinrite.htm

It checks the drive surface pretty well, using DOS.

 

[Ian East]

You might look into Spinrite.

Spinrite looks pretty good. I'll definitely have to check it out.
Thanks for the link.

 

[Ian East]

"Dependent"?
What exactly is this OS image?

It's an os image for a appliance. The OS is proprietary and we do not
have access to it outside the normal user interface (I assume it's
running QNX or some embedded systems OS but I'm not sure). We have
found that we can dd an image from a good disk and dump it to a new
disk successfully, but when using a different type of disk it fails.

Fair enough, but given you already feel this potential
exists, is it prudent to deploy systems with these old
drives?

When faced with the decision of choosing between bad or worse,
sometimes we must make these decisions. It's only a temporary
solution until the hardware is replaced.

Seems like a liability to me, either of data or a lawsuit or
??? Depends on the use.

It's risk that has to be managed. Keeping mirrors, diligent backups,
and having hot swap hardware can help a bit.

No, that's reckless and risks your life, maybe others' as
well. Wouldn't someone poor enough (or with different
priorities) that they have a car worth less than a set of
tires be too poor/whatever to be buying a new car at the
beginning of the next year? If there is budget for a new
car, tires costing little more than one, "maybe" two month's
payments are not such a burden. Come to think of it the
insurance alone on the new car is going to cost enough that
new tires are still a better solution if they're only used
for a month or two, which would help to sell the car as
well, or rather, might make it sellable at all. I've gone
off on a tangent but I still disagree completely with your
concept of testing something old enough that it's inevitable
that it fail sooner rather than later.

Perhaps the analogy of bald tires is a bit off track as this is
certainly not a matter of life and death. Trying to extend the life
of an aging piece of hardware, car, or human is not that uncommon of a
thing.

Unfortunately, in the real world, many companies (often auto
manufacturers ironically) are in a position of making life vs
economics decisions on a daily basis. Many times they favor economics
for a short gain only to suffer huge long term consequences. Somehow
they seem to forget their lesson when quarterly earnings reports are
due.

Then how do you expect to keep that proprietary OS image?
Write testing generally is done to a blank (or soon-to-be)
drive.

If this OS image is valuable I'd think you should be keeping
an archival backup of it, and it would be even more prudent
given the propensity for failure even while doing the
testing. If you make such a backup with Driveimage or Ghost
you should then have either of them read and make a backup
image (while there is still greatest chance for intact data)
then restore that image using the write-verify option the
programs offer.

The idea is to erase the disk. Qualify it. Then write the OS image
back to the disks that pass. Once the image is loaded, the new
configuration for the appliance can be uploaded quite easily.

 

[CBFalconer]

What does "re-qualification" mean? I believe you can run a program to
say how many sectors are good or bad. And you can verify that there
are no errors during an X-hour run. But when I purchase a drive, I'm
looking for an indication that the drive will continue to serve me for
some number of years in the future. I want to know how many hours are
on the bearings, whether it has been operated in a high temperature
environment that might have left the electronic components dried out.

None of which you can really tell without the cooperation of the
previous owners. However Steve Gibson (grc.com, I believe) has
software to do all sorts of delicate tests on hds. He can detect
the degree of margin available. I think the program is called
Spinrite.

 

[kony]

It's an os image for a appliance. The OS is proprietary and we do not
have access to it outside the normal user interface (I assume it's
running QNX or some embedded systems OS but I'm not sure). We have
found that we can dd an image from a good disk and dump it to a new
disk successfully, but when using a different type of disk it fails.

IMO, it might be worthwhile to focus more on why this won't
transfer to another disk and attempt to get to it run from
flash.

 

[Joe Morris]

I have many hard disks that I must re-qualify (they are old and can
not easily be replaced). Does anyone know of a good way of
requalifying IDE hard disks on an x86 system? The OS doesn't matter
as I can use Linux, Windows, Solaris x86, etc... The test should at
least write the entire surface of the disk, read the entire surface of
the disk, do random seek tests, and "exercise" the disk for 12 hours
or so. Does anyone have any suggestions? I've tried some of the
various benchmarking programs like bonnie++, iozone, etc... but have
not quite found something that suitable specifically for
re-qualification.

It's been more years than I care to remember since I last looked
at it, but at one point there was a disk test utility packaged with
the old Norton Utilities.

And the old pre-IDE controllers usually had a low-level formatter
built into the controller's BIOS (IIRC entered at <segment>:5 using
the DEBUG program). If the disks are so old that they have some
physical characteristic (the antique Compaq thinline disks, perhaps?)
it's possible that the controller you're using might have such a
feature.

I do have to ask, however: what is there about the disks that
precludes physical replacement with newer drives?

Joe Morris

 

[Joe Morris]

It's an os image for a appliance. The OS is proprietary and we do not
have access to it outside the normal user interface (I assume it's
running QNX or some embedded systems OS but I'm not sure). We have
found that we can dd an image from a good disk and dump it to a new
disk successfully, but when using a different type of disk it fails.

H'mmm...have you considered talking to the DiskManager folk and asking
how much it would cost to get a custom-built shim that would sit between
the (real) disk geometry and the OS? Recall the popularity of DM when
the 508 megabyte boundary was passed in the mid-1990s: it remapped
the disk geometry (few heads, too many cylinders) into a virtual
geometry of lots of heads and sufficiently few cylinders to be
acceptable to the older DOS systems.

This probably won't be pocket change if you try this route, but
depending on your situation it might free you from dependency on
the specific geometry of the disks you have.

Joe Morris

 

[Jiminy]

Try SpinRite, Steve Gibson's program.

http://www.grc.com/sr/spinrite.htm

It checks the drive surface pretty well, using DOS.

....and takes more than 18 hrs to scan a 60Gb drive!

best,

Jiminy

 

[VWWall]

...and takes more than 18 hrs to scan a 60Gb drive!

Of course it does! You can't examine something without looking at it!
The program tries each sector in several ways, looking for weakness. No
program can examine drive sectors faster than the drive can access them.

This is in the category of "no free lunch".

 

[William W. Plummer]

Of course it does! You can't examine something without looking at it!
The program tries each sector in several ways, looking for weakness. No
program can examine drive sectors faster than the drive can access them.

This is in the category of "no free lunch".

There will be a great difference between accessing sectors in the "best"
order where a sector transfer is started just as it comes under the
heads, and the "worst" order, where a transfer is started for the sector
just passed. In the latter case, you will have to wait for a whole
revolution of the disk on every sector. The former case will operate as
fast as the physical disk will permit.

To make this happen, the program must know a lot about the physical
layout of the disk. That is largely hidden these days, but some clever
disk guys know how to do it.