When is a defective drive defective.

[jurgen36]

About 4 years ago one of my Seagate 20 Gigabyte drives developed numerous bad sectors. I used the Seagate test program and it told me that the drive is us and has to be replaced.

However I wanted to see what is really going on and zero filled the drive. Then I reformatted the thing with fat32 and used it only as a backup for non essential programs.

Well that was 4 years ago, even today the drive is in perfect conditions and outlived many of the later drives.

So the question is when is a drive really faulty, regardless what the test programs say. I would suggest before disregarding an older drive to zero fill it and then format it. You might be surprised with the results.
Regards

Jurgen

 

[BNPrchr]

When you wrote "I reformatted the thing with fat32...today the drive is in perfect conditions", I assumed it was originally an NTFS volume. I have had several NTFS-formatted drives "fail", and have seen many "corruptions" of NTFS metadata areas reported by "chkdsk /F" (or "chkdsk /R", depending on the boot context), [thank you, Micro$oft]. Since the various FAT filesystems do not place anywhere the metadata processing overload on the disk controller(s) and/or processor(s) as do NTFS filesystems, I simply assume that FATnn _can_ be more reliable than NTFS under certain circumstances.

While Micro$oft claims that NTFS is "more robust", there is a question of availability of programs/data, is there not?

 

[jurgen36]

:BNPrchr
Yes you are correct it was originally in NTFS format. Now I have all my drives including 1 terrabyte drives in Fat 32. Seems to be much more reliable. I also tested the speed of Fat32 versus NTFS and found that in most cases Fat32 is faster and seems to be much more reliable. Also in case of corruption the file recovery seems to work much better.

Regards

Jurgen

 

[cdogg]

Many tests in the past were performed to measure the difference in speed. The consensus is that FAT 32 was only quicker than NTFS on smaller partitions less than 32GB in size. They are fairly equal after that mark until sizes reach 70-80GB and beyond, in which NTFS grabs the advantage. But that's a whole different rant to take on!

As for your Seagate drive...

I'd say it was a case of good luck. Bad sectors can be caused by a number of reasons: shock to the drive while it is spinning/reading, mechanical failure, software write failure (BSOD), etc...

Obviously, there are situations where a zero format will end up resolving the issue, especially if the sectors aren't permanently damaged. When the sectors are unusable and can't be repaired, there are still times when that was caused by something other than mechanical failure (such as shock), meaning the heads and motors are fine and will continue to read the remaining sectors without problems. This was most likely the case you experienced.

Generally, however, hard drive tests look at a lot of different factors measured by S.M.A.R.T. taking seek times and bad sectors into consideration. I'm not sure about the test you ran, but the HDD test I use from HP has been very accurate in predicting imminent drive failure.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[jurgen36]

:cdogg

Yes I do agree with most of what you said. However with drive speeds Fat32 versus NTFS I tend to disagree. Until my retirement last year I was head of the computer science department of a major Thai University. We did run a multitude of speed tests in our labs and they practically always were in favor of the Fat32 drives, regardless of drive size. Unfortunately as I am no longer in Thailand I can not publish our extensive test results.

Yes it might have been good luck, I normally do not use small 20 Gbyte drives, but I wanted to find out how long the drive will last. The test I was using was the standard Seagate test which is used before the drive was returned to the company. It showed hundreds of bad sectors. Could be that they were caused by a power failure during a write cycle. I would not really know. However after a zero fill and reformat no bad sectors showed up. This could be explained by the drive replacing the bad sectors with spare unused sectors. Again I was too busy to really investigate the issue.

All I wanted to point out that a zero fill might make a faulty drive usable again.

Anyway best of regards.

Jurgen

 

[cdogg]

[blue]We did run a multitude of speed tests in our labs and they practically always were in favor of the Fat32 drives, regardless of drive size. Unfortunately as I am no longer in Thailand I can not publish our extensive test results.[/blue]"

I can accept that. I agree that there are a lot of factors involved (disk fragmentation, drive buffer, size of files, etc.) that would impact whether or not FAT32 outperforms NTFS in some situations. But the fact alone that NTFS has built-in fault tolerance when handling moves/writes is enough to persuade me to ditch FAT32. In addition, FAT32 is more prone to "cause" fragmentation and "suffer" as a result taking a hit in performance.

~cdogg
"Insanity: doing the same thing over and over again and expecting different results." - Einstein
[tab][navy]For posting policies, click [/navy]here.

 

[edfair]

I don't deal with NTFS much so my comments hold more with FAT16 and FAT32 drives.
Over the years I've come to the same conclusions regarding drives that have failed in operation, that they sometimes are not really bad, but confused.
I normally mark the removed drives as failed and throw them in a box for later inspection. When the time comes they go on a system with a different controller and a shorter single connector cable for the duration of a zero fill, fdisk, and format. Those that originally failed to ID to the BIOS generally fail the same way and are discarded. The rest often times come back to life.
Some of this parallels the issues I've had over the years where customer machines come back to life with a change of cable.
I'm not that much into drive diagnostics but in searching out the issues when I was using more IDE drives somebody pointed out that the data transfer between drive and channel was unterminated and subject to failures because of it. I never put a scope on any so can't properly evaluate that assertion.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[acewarlock]

Quote
"I'm not that much into drive diagnostics but in searching out the issues when I was using more IDE drives somebody pointed out that the data transfer between drive and channel was unterminated and subject to failures because of it. "

Unterminated usually pretains to SCSI HD and not IDE. If you don't Terminate a SCSI HD you will get errors sometimes.


This is a Signature and not part of the answer, it appears on every reply.

This is an Analogy so don't take it personally as some have.

Why change the engine if all you need is to change the spark plugs.

 

[edfair]

Agreed to the termination issues with SCSI. Also MFM and all others except IDE. And floppy drives. But haven't had a reason to check on what steps have been taken in the last 10 years with floppies.
I consider any drive sensitive to ribbon cable length as a probable candidate for noise issues.

 

[acewarlock]

SCSI is the only type that ever required a Terminator on the cable.


This is a Signature and not part of the answer, it appears on every reply.

This is an Analogy so don't take it personally as some have.

Why change the engine if all you need is to change the spark plugs.

 

[edfair]

If I recall correctly there were terminators on MFM & Rll drives. I don't remember ever tracing the lines that were terminated but evidently the resistors were there for some purpose and were needed. It has probably been 15 years since I've touched either type of drive and I can't recall the symptoms of a missing terminator.
SCSI has a roughly midpoint operating voltage split 220 ohms and 330 ohms between 5v and 0v but the predecessor SASI controllers didn't have them. So I assume that the midpoint aids in getting the correct signals over a faster bus.

 

[acewarlock]

I have been repairing and selling used HD for 20 years and I have never seen a Terminator on any drive type but SCSI HD's


This is a Signature and not part of the answer, it appears on every reply.

This is an Analogy so don't take it personally as some have.

Why change the engine if all you need is to change the spark plugs.

 

[acewarlock]

With MFM and RLL HD's each drive needed a connector that was reversed from the other drive, with IDE they are on the same cable and use jumper plugs.


This is a Signature and not part of the answer, it appears on every reply.

This is an Analogy so don't take it personally as some have.

Why change the engine if all you need is to change the spark plugs.

 

[edfair]

try a google search "terminator mfm". That will pick up some info for you.

 

[acewarlock]

OK I stand corrected, I forgot about those little buggers on the drive. They are so small compaired to the one's on SCSI drives that you forget about them.

On a lot of older( i.e. Full-Height) drives, the term pack sits on the drive's controller board near the connectors. Often,it's some shade of yellow, and has 14 or 16 pins and looks like any regular (yellow) chip, even though it's really not, it's a grid of resistors which dampen the signals on the floppy( or hard drive) cable, so that on reflection( when the signal hit's the "end" of the cable) it get's dampened so that it doesn't interfere with 'proper' operation.




This is a Signature and not part of the answer, it appears on every reply.

This is an Analogy so don't take it personally as some have.

Why change the engine if all you need is to change the spark plugs.