Corrupt NTFS filesystem

[Rod Speed]

I ran chkdsk. The corruption problem occurs seemingly at
random. The longer I go between reboots, the more likely
corruption is. I have gone as long as a week with no
problems only to have it back again on the next day.

That sounds like some fundamental hardware problem
like the motherboard doesnt like the ram much.

I'd run memtest86 overnight every night because it
looks rather like the problem is rather intermittent.

You don't seem to understand the exact nature of this problem.

You never did say it very clearly.

I have to reboot to run chkdsk, and that is when the
corruption shows up. It does not show up unless I reboot.

Likely that is just because it isnt corrupted enough to affect normal ops.

Ironically, if the corruption occurs, Win2K runs chkdsk for me at boot time.

Thats rather odd. That may be a coincidence tho, just that you are
only aware of the corruption when chkdsk is run and when 2K chooses
to run it, presumably because it decides that it wasnt shut down
properly, you see the corruption just because chkdsk has been run.

Sometimes the corruption is so extensive that I get a BSOD. That's when
I have to mount the disk as D: and run chkdsk from another boot disk.

Thats even odder.

 

[kony]

You said fix the problem. If you know how to make 98 manage memory
properly and not crash regularly, occasionally leaving FAT problems
behind, I'm all ears!

Non-applicable, '98 doesn't allow the choice of NTFS.

I do have systems that run FAT32 on '98 though, and can't
recall the last time any files were lost.

 

[kony]

Which 3 keys - my Registry has 5 hive keys.

I must have missed something along the way, because I do not see how
exporting hive keys is going to do any good. If the Registry is
screwed up, then exporting hive keys is going to carry the screwed up
parts along.

I mentioned specific keys.

"IF" the registry is screwed up and that's the problem, then
reproducing that problem by only copying some reg keys will
at least help to isolate it.

I can see you are still not understanding the situation- the
only thing keeping you from having a properly working
system is _you_. You are trying to think, or argue, instead
of _doing_. Taking the approach you have, no problems
anyone has would be solved, but clearly problems do get
solved.

 

[meow2222]

Non-applicable, '98 doesn't allow the choice of NTFS.

I've put ntfs on 98, courtesy of Mr Russinovich. Read only
unfortunately.

I do have systems that run FAT32 on '98 though, and can't
recall the last time any files were lost.

Lucky you. Maybe mostly running older drives comes into it too, but
I've seen 98 mangle a FAT on several occasion.


NT

 

[Citizen Bob]

I can see you are still not understanding the situation- the
only thing keeping you from having a properly working
system is _you_. You are trying to think, or argue, instead
of _doing_. Taking the approach you have, no problems
anyone has would be solved, but clearly problems do get
solved.

Thanks for the lecture.


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Citizen Bob]

That sounds like some fundamental hardware problem
like the motherboard doesnt like the ram much.

I'd run memtest86 overnight every night because it
looks rather like the problem is rather intermittent.

I have run it but not every night.

You never did say it very clearly.

I did many times earlier. This thread has been recurring for several
months now. Apparently you were not participating on this particular
forum back then.

Likely that is just because it isnt corrupted enough to affect normal ops.

It is pretty corrupted when it BSODs. CHKDSK fusses about all sorts of
things. I posted a file containing all the crap CHKDSK fixed and it is
several pages long. Next time I get a BSOD, I will post it again.

Thats rather odd. That may be a coincidence tho, just that you are
only aware of the corruption when chkdsk is run and when 2K chooses
to run it, presumably because it decides that it wasnt shut down
properly, you see the corruption just because chkdsk has been run.

There are entries in the Event Viewer - "corrupt ntfs volume, run
chkdsk". Not very helpful. Nothing else seems to be wrong in EV.

After a few days of running the "new" NTFS partition, the one made
from the earlier FAT32, I finally got a corrupt disk. When it was
booting it ran CHKDSK on its own. The only apparent problem was "blank
space in the MFT".


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Citizen Bob]

All you have to do is make a clean
installation, export the appropriate registry keys and copy
over the installation folders and shortcuts for the start
menu.

I still do not know what Registry keys you are referring to.

It is very likely that the corrupt ntfs filesystem is coming from a
Registry key. So how do I prevent it from contaminating the new
install?


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Rod Speed]

I have run it but not every night.

I'd run it every night for a few weeks, because it
looks like it might well be an intermittent problem.

I did many times earlier.

Nope, you never spelt it out that clearly before.

It wasnt initially clear that reformat to FAT32 had fixed just
SOME of the problems and not the duplicate entrys either.

This thread has been recurring for several months now.
Apparently you were not participating on this particular
forum back then.

Its silly to rely memory of stuff that far back even if I was.
There's been a few more than just you with problems over that time.

It is pretty corrupted when it BSODs.

Sure, I meant when it doesnt, just runs chkdsk itself at boot time.

CHKDSK fusses about all sorts of things. I posted a file
containing all the crap CHKDSK fixed and it is several
pages long. Next time I get a BSOD, I will post it again.

Dont bother, I can find it using groups.google
and the detail really doesnt matter much.

There are entries in the Event Viewer - "corrupt ntfs volume, run chkdsk".

And that is something you should have said earlier too.

Not very helpful.

Dunno, how soon those show up might well be useful info
about how soon you start seeing corruption after a cleanup.

Nothing else seems to be wrong in EV.

After a few days of running the "new" NTFS partition, the
one made from the earlier FAT32, I finally got a corrupt
disk. When it was booting it ran CHKDSK on its own.

When did the first EV entry show up in that sequence ?

The only apparent problem was "blank space in the MFT".

Clearly something is screwing with the directory structures.

I'd personally do a clean install of 2K on a spare hard drive
and run that for days to see if that gets corrupted too. Basically
to see if the problem is with the hardware or a ****ed 2K install.

It'd be hilarious if it turns out to be those removable drive bays that are doing it.

 

[kony]

Lucky you. Maybe mostly running older drives comes into it too, but
I've seen 98 mangle a FAT on several occasion.

Nope not old drives, and not luck. Though win98 had several
issues people were frequently glad to be rid of moving to NT
based OS, losing files because of FAT32 just wasn't one of
them.

 

[kony]

Thanks for the lecture.

Is it not true?

If it were more useful to continue down the path you're
taking, I would've- but how many times and posts have
transpired already? I'm posting what I consider the best
course to rectify the problem, that course being to do...
already plenty of though on it but that didn't help so it's
time to shift gears.

 

[kony]

I still do not know what Registry keys you are referring to.

I posted them. HKCU-software, HKLM- software,
HK-Classes-Root. The abbreviations are obvious enough when
looking at regedit.

You have not yet done the clean install though? Trying to
look at the whole process in one lump is always more
difficult than taking things one step at a time and it's not
time to merge the registry if you dont' have a known good
clean 2KSP4/patched installation up yet.

It is very likely that the corrupt ntfs filesystem is coming from a
Registry key. So how do I prevent it from contaminating the new
install?

You dont' need to. You get system working before reg keys
are added, make a backup. Make sure it works right, no
phantom double drive volume entries or anything like that
and if adding the registry entries causes the problem you
have then isolated it to a registry setting which is a GOOD
thing, more than you have to go on so far.

It shouldn't be though, you're only merging the software
keys, not the entire registry, and ideally weeding out
software you dont' even have installed anymore before
exporting it, but if you want to save time and not weed that
stuff out, that's ok too- it'll just be a little clutter
opposed to a lot of it and the problem.

The main thing to remember is you are not trying to do all
of this (registry AND files) at once. Do one, and verify
the system works properly. Same goes for comparing the
driver folders- comparison should be done before adding
anything from the old installation.

There's no point in typing anything more if you're not going
to get the system into a state where it could be applied.

 

[CBFalconer]

On 27 Oct 2006 01:22:05 -0700, (e-mail address removed) wrote:
.... snip ...

Nope not old drives, and not luck. Though win98 had several
issues people were frequently glad to be rid of moving to NT
based OS, losing files because of FAT32 just wasn't one of
them.

I guess you haven't got him plonked for obscenity.

 

[Citizen Bob]

Dont bother, I can find it using groups.google
and the detail really doesnt matter much.

I finally got a BSOD this morning, but I forgot to redirect the
output. It was the same sort of thing as it always is with a BSOD. And
this was only 24 hours after yesterday's boottime automatic chkdsk.
IOW it appears that the corruption is beginning to occur more often.

I do not have enough data to state this with certaintym but I have
noticed that if I defrag with PerfectDisk, both online and offline,
that I can go a lot longer before corruptions. After a defrag I can
get as much as a week before corruption. Then they come more often
until I defrag again.

One hypothesis is that those dual devices are confusing the MFT and it
gets worse with time until it finally crashed the system upon boot.
Eventually it gets so bad it BSODs on boot.

When did the first EV entry show up in that sequence ?

Here is the usual sequence. I look in EV and see nothing wrong. I
clean it out so there are no entries. I shut down and reboot. Most of
the time when something goes wrong I get an automatic chkdsk, and
about 10% of those times I get a BSOD instead. After fixing the disk I
look in EV and that's when I see the Corrupt NTFS Volume message. It
always appears twice in succession at the time of the reboot.

Clearly something is screwing with the directory structures.

I'd personally do a clean install of 2K on a spare hard drive
and run that for days to see if that gets corrupted too. Basically
to see if the problem is with the hardware or a ****ed 2K install.

You still do not appreciate the situation I am in. I cannot run an
empty Win2K. I have to have applications and that's what I do not want
to do - reinstall applications on a new Win2K.

If I had another machine I could do it although I would not be
exercising the applications since there would not be any.

There is always the possibility that a rogue application I am running
is causing the problem. If I could get a stable baseline for the
system, then I could keep track of what I run each day.

It'd be hilarious if it turns out to be those removable drive bays that are doing it.

I have no evidence to support that. If the removable bays were causing
the trouble then I would see the corruption happening at times other
than reboot. Also, I would see it happening to the backup disk in D:


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Citizen Bob]

I posted them. HKCU-software, HKLM- software,
HK-Classes-Root.

I never saw that post. You did not post any Registry keys like the
ones above on this forum that I can find.

I ran a Google Groups Advanced search on this forum with "HKCU" as the
keyword. I found only this exact post - the one I am replying to right
here - and one other that had nothing to do with this discussion.

Either your computer is broken or you are posting to another group or
you are hallucinating. But don't take my word for it - check it out
yourself on Google. If you find it, then I am hallucinating.

It shouldn't be though, you're only merging the software
keys, not the entire registry, and ideally weeding out
software you dont' even have installed anymore before
exporting it, but if you want to save time and not weed that
stuff out, that's ok too- it'll just be a little clutter
opposed to a lot of it and the problem.

If those are the only keys I have to deal with, then I can weed them
out in a text editor or a Registry editor.

There's no point in typing anything more if you're not going
to get the system into a state where it could be applied.

Now that I know what keys to export, I can guage the size of the
project. It looks doable now, so I will put it on my calendar. I am
currently involved in reprogramming my remote control using JP1 hacks
for new DVDs I bought but there are no codes available.

It seems all I am doing nowadays is fixing things.


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Rod Speed]

I finally got a BSOD this morning, but I forgot to redirect the
output. It was the same sort of thing as it always is with a BSOD. And
this was only 24 hours after yesterday's boottime automatic chkdsk.
IOW it appears that the corruption is beginning to occur more often.

I do not have enough data to state this with certaintym
but I have noticed that if I defrag with PerfectDisk, both
online and offline, that I can go a lot longer before
corruptions. After a defrag I can get as much as a week before
corruption. Then they come more often until I defrag again.

I bet that is an illusion, and if it isnt, likely its just because a defrag
minimises the changes that are needed to the MFT due to normal
file activity, and so it gets corrupted slower just due to that.

One hypothesis is that those dual devices are confusing the MFT

Cant be that because access will always
be via one of the devices with normal ops.

and it gets worse with time until it finally crashed the system
upon boot. Eventually it gets so bad it BSODs on boot.

Here is the usual sequence. I look in EV and see nothing wrong. I
clean it out so there are no entries. I shut down and reboot. Most of
the time when something goes wrong I get an automatic chkdsk, and
about 10% of those times I get a BSOD instead. After fixing the disk I
look in EV and that's when I see the Corrupt NTFS Volume message.
It always appears twice in succession at the time of the reboot.

You still do not appreciate the situation I am in.
Wrong.

I cannot run an empty Win2K.

Corse you can for a TEST.

I have to have applications and that's what I do not
want to do - reinstall applications on a new Win2K.

I'm talking about a TEST, not a new config.

If I had another machine I could do it

You dont need another machine, just another hard drive for the test.

although I would not be exercising the
applications since there would not be any.

Even you should be able to install just a couple of apps for the TEST.

There is always the possibility that a rogue
application I am running is causing the problem.

Unlikely since few apps directly manipulate the MFT, they leave that to the OS.

That is however why I suggested not using Perfect Disk
for a while, in case it that thats corrupting the MFT.

If I could get a stable baseline for the system,
then I could keep track of what I run each day.

You dont need to have a stable baseline to do that.

I have no evidence to support that.

Yes, but it would be worth trying the boot drive directly connected
instead of in the removable drive bay to eliminate that possibility since
we do know that you can get that result with removable drive bays.

If the removable bays were causing the trouble then I would
see the corruption happening at times other than reboot.

You dont know that you dont get it at times other than a reboot.

And even if it does only happen on a reboot, that proves nothing
about what is doing that. If it only happens on a reboot, it cant be
an app doing it, it has to be the OS or the hardware.

Also, I would see it happening to the backup disk in D:

Not necessarily.

Anyone with a clue would try the boot drive directly connected and see if
the corruption still occurs. Bet you'll find a reason not to do that obvious test.

 

[Citizen Bob]

Wrong.

Do you depend on your computer every day to make money? Or do you just
use it for recreation?

Corse you can for a TEST.

In order to test the new install, I would have to run it full time for
several days. What am I supposed to do about all the apps I normally
run in the course of a day? I can't just abandon my routine for a test
- I need to run the apps every weekday.

I'm talking about a TEST, not a new config.

If I do not install enough apps then I can't run the things I need to
run.

You dont need another machine, just another hard drive for the test.

How am I going to run two versions of Win2K on two separate partitions
at the same time?

Even you should be able to install just a couple of apps for the TEST.

You don't realize that I would have to install more than just a couple
apps.

Unlikely since few apps directly manipulate the MFT, they leave that to the OS.

What about those that don't? I did say "rogue applocation".

I am just guessing so let's assume you are right and that is not the
problem.

That is however why I suggested not using Perfect Disk
for a while, in case it that thats corrupting the MFT.

PerfectDisk is very new - only a month or two old. This problem has
been going on for a year.

You dont need to have a stable baseline to do that.

Without a stable baseline, I can't be sure that any one variable is
contibuting.

Yes, but it would be worth trying the boot drive directly connected
instead of in the removable drive bay to eliminate that possibility since
we do know that you can get that result with removable drive bays.

If I connect it directly, then I have to mount it inside the machine.
Then when it gets corrupt I can't make it the D: drive without taking
it out.

Actually that is not quite true. My BIOS allows me to choose which
disk is the boot disk, so if I wanted the hard mount to be D:, I make
the disk in the tray the C:.

I may do just what you suggest - make one of the disks a hard mount
bypassing the tray. I may have a cabling problem because of the
location of the internal vs removable bay. However I have some
brackets to turn a 5 1/4: iinto a 3 1/2" so I could keep the disk in
the same position in the case.

You dont know that you dont get it at times other than a reboot.

U check EV all the time and have never seen it. However it may be that
it is not being detected except at boot. I can't run chkdsk whenever I
want - it must be run at boot. Is there some other diagnostic that can
detect a corrupt NTFS volume that I could schedule to run periodically
while Win2K is running?

And even if it does only happen on a reboot, that proves nothing
about what is doing that. If it only happens on a reboot, it cant be
an app doing it, it has to be the OS or the hardware.

Good point, if it is only happening at boot.

Not necessarily.

Anyone with a clue would try the boot drive directly connected and see if
the corruption still occurs.

I had a good reason not to for which I found a good workaround.

Bet you'll find a reason not to do that obvious test.

Your lost your bet. If I don't want to do something it is not because
I am lazy or obstinate - it's because I believe I have a good reason
not to. Of course I could be wrong, but until I realize that, I am not
going to go off into the weeds.


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Citizen Bob]

That is however why I suggested not using Perfect Disk
for a while, in case it that thats corrupting the MFT.

PerfectDisk is very new - only a month or two old. This problem has
been going on for a year.

Something I just thought of. When I first started using PerfectDisk,
every once in a while it would cause the same corruption problem. I
knew because before I ran PD, I would reboot to make sure the volume
was not corrupt and then I would create a clone backup. Then I would
reboot and run PD and then reboot to see if it corrupted the disk,
Sure enough, a couple times it did and I had to use either the clone I
just made to recover or run chkdsk.

Speaking of clones I think I mentioned this but sometimes you may not
have picked up on it. If I put the clone in the D: without changing
the signature with Win98SE fdisk /mbr, it will always BSOD. That's
because Win2K tried to mount the same device to two disks with
identical signatures. If I use the trick of Win98SE fdisk /mbr on the
D: disk, then I do not get the BSOD. Of course Win2K prompts me to
reboot because it has found a "new device".

So let's imaging the scenario where I have a clone in archive which I
use as the boot disk when the original disk gets corrupted. I mount
this archived clone as the boot disk and mount the bad disk as D: so I
can run chkdsk d: /f on it. Since I changed the signature on the bad
disk to prevent the device conflict and BSOD, and if I don't reboot to
satisfy Win2K's request for a new device, then chkdsk will screw up
the bad disk and it is not recoverable. IOW, it can't even be mounted
any more.


--

"Nothing in the world can take the place of perseverence. Talent
will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb. Education
will not; the world is full of educated derelicts. Persistence and
determination alone are omnipotent."
--Calvin Coolidge

 

[Rod Speed]

Do you depend on your computer every day to
make money? Or do you just use it for recreation?

Irrelevant to what is possible for YOU to do for a TEST.

In order to test the new install, I would
have to run it full time for several days.
Wrong.

What am I supposed to do about all the apps I normally
run in the course of a day? I can't just abandon my routine
for a test - I need to run the apps every weekday.

I doubt you actually run all that many of them every weekday
and if you do, you can certainly do the other test, try with the
drive directly connected instead of in a removable drive bay.

If I do not install enough apps then I can't run the things I need to run.

I doubt that involves all that many apps, and if it does,
you can certainly do the other test, try with the drive
directly connected instead of in a removable drive bay.

How am I going to run two versions of Win2K
on two separate partitions at the same time?

You dont have to run them at the same time.

You don't realize that I would have to install more than just a couple apps.

Then you can obviously install what you do need to run,
and you can certainly do the other test, try with the drive
directly connected instead of in a removable drive bay.

What about those that don't?

Its unlikely that you have anything more than Perfect
Disk which doesnt which you run every weekday.

I did say "rogue applocation".

I am just guessing so let's assume you
are right and that is not the problem.

No need to assume anything, its easy to TEST that proposition.

PerfectDisk is very new - only a month or two old.
This problem has been going on for a year.

OK, you never made that clear till now.

Without a stable baseline, I can't be
sure that any one variable is contibuting.

Wrong, that is just one way to test.

If I connect it directly, then I have to mount it inside the machine.
Then when it gets corrupt I can't make it the D: drive without taking it out.

Hardly the end of civilisation as we know it. If it does end up corrupted
when directly connected you would presumably want to put it back in
the removable bay anyway so you can fix it once you have done that.

Actually that is not quite true. My BIOS allows me to choose
which disk is the boot disk, so if I wanted the hard mount to
be D:, I make the disk in the tray the C:.

See above.

I may do just what you suggest - make one of the disks a
hard mount bypassing the tray. I may have a cabling problem
because of the location of the internal vs removable bay.

Unlikely given that its easier to cable an internal than a removable bay.

And even if you did need to get another cable for the test, that is
well worth doing because its very likely what is corrupting the MFT.

However I have some brackets to turn a 5 1/4: iinto a 3 1/2"
so I could keep the disk in the same position in the case.

I'd make sure it was a proper legal ATA cable too,
a proper 80 wire flat ribbon cable of legal length. No
point in doing the test with a known non standard cable.

U check EV all the time and have never seen it.

All that means is that the OS hasnt noticed it until boot time.

However it may be that it is not being detected except at boot.
Precisely.

I can't run chkdsk whenever I want - it must be run at boot.

No it doesnt need to to just CHECK for corruption,
only for FIXING any corruption seen.

Is there some other diagnostic that can detect a corrupt NTFS
volume that I could schedule to run periodically while Win2K is running?

You dont need one, chkdsk can do that fine.

Good point, if it is only happening at boot.

I had a good reason not to

No you didnt.

for which I found a good workaround.

You dont need a workaround. If the directly connected drive
still gets corrupted, you'd presumably want to put it back in the
removable drive bay so you can repair it after you have done that.

Your lost your bet.

Not yet, you havent actually done that test yet.

If I don't want to do something it is not because I am lazy or obstinate

That remains to be seen. You've got one hell
of a capacity for refusing to do the obvious tests.

- it's because I believe I have a good reason not to.

Thats just the excuse for the bone headedness.

Of course I could be wrong, but until I realize
that, I am not going to go off into the weeds.

We'll see...

 

[Rod Speed]

PerfectDisk is very new - only a month or two old.
This problem has been going on for a year.

Like I said, you've only just said that.

Something I just thought of. When I first started using PerfectDisk,
every once in a while it would cause the same corruption problem.
I knew because before I ran PD, I would reboot to make sure the
volume was not corrupt and then I would create a clone backup.
Then I would reboot and run PD and then reboot to see if it
corrupted the disk, Sure enough, a couple times it did and I
had to use either the clone I just made to recover or run chkdsk.

That would seem to indicate that its actually disk activity that
produces the corruption, supporting the possibility that is just
something as basic as the removable drive tray thats the problem.

There clearly isnt any other app involved
when an ImPerfect Disk run corrupts the drive.

Bet its the removable drive bay or the cable.

Speaking of clones I think I mentioned this but sometimes
you may not have picked up on it. If I put the clone in the
D: without changing the signature with Win98SE fdisk /mbr,
it will always BSOD. That's because Win2K tried to mount
the same device to two disks with identical signatures.

I dont believe that claim about 2K, essentially because clones work
fine for others without that abortion involving Win98SE fdisk /mbr

What are you doing the cloning with again ?

If I use the trick of Win98SE fdisk /mbr on the D: disk,
then I do not get the BSOD. Of course Win2K prompts
me to reboot because it has found a "new device".

So let's imaging the scenario where I have a clone in archive

Not clear what you mean by that last.

which I use as the boot disk when the original
disk gets corrupted. I mount this archived clone

Or that either.

as the boot disk and mount the bad disk as D: so I can run
chkdsk d: /f on it. Since I changed the signature on the bad
disk to prevent the device conflict and BSOD, and if I don't
reboot to satisfy Win2K's request for a new device, then
chkdsk will screw up the bad disk and it is not recoverable.
IOW, it can't even be mounted any more.

Cant really understand what you mean
config wise with those 'archive' comments.

 

[Rod Speed]

Like I said, you've only just said that.


That would seem to indicate that its actually disk activity that
produces the corruption, supporting the possibility that is just
something as basic as the removable drive tray thats the problem.

There clearly isnt any other app involved
when an ImPerfect Disk run corrupts the drive.

Bet its the removable drive bay or the cable.

And ImPerfect Disk is the obvious thing to use when testing
for corruption, no need to run the normal 2K install for days etc.