no driver

New HDD burn-in routines?

Recommended Posts

I think that stress-testing/burn-in should be separated from a write to all sectors once.

If I buy a new drive the first thing I do is a zero fill and when it shows reallocated sectors I return that drive as a new drive should not have the need to remap IMO.

Torturing the drive endlessly for days may not be very helpful, but I don't trust a drive as long as it hasn't written all of its sectors at least once.

Share this post


Link to post
Share on other sites
6_6_6, I would be interested to know how many of the 100 drives passed steps 1 and 2, but failed steps 3-8 before handling any real data. (When I put a disk into service, I only do steps 1 and 2.)

I think i had about 7-8 drives failed (somewhere along these steps) and had to be returned. It doesn't matter at which point they failed, coz i would be running most of the tests to see if drive can be healed. I never had a DOA.

It has been widely reported that the failure rate of HDDs follows a "bath tub" curve. It is not fully known how much of the "near wall" of the bath tub curve (i.e., the relatively high failure rate in disk drives early on) is caused by shipping/handling damage.

And since 6_6_6 takes drives out of operation after 2 years, it is of no consequence that the "far wall" of the bath tub curve may be reached sooner after performing a stress-test.)

The reason i take them offline for intermittent use is because manufacturers recommend 8 hours per day usage for non-enterprise drives... and those drives have been running 24/7 for a year or two. So essentially i have been running these drives 3-6 years as per manufacturers' criteria.

Share this post


Link to post
Share on other sites

Okay, 4 months more passed and i still have no drives failed that passed a stress test (including the one with disasterous graphs i posted in the first picture).

During this time I had 2 defective drives. One WD 1TB which was okay to boot with (SMART, read-write, etc) but gone haywire after a run of SMART long-test.

Another one was a Samsung F1 1TB that my cousin brought over. 2 months old and he said he was having occasional BSODs from the very beginning. All was okay with the drive initially, but developed so many UNCs and Off-Track errors after a run of manufacturer's zero-fill that i did not bother to test anything more.

So if you are any wiser, the absolute minimum you must do prior to commiting any data to a drive is:

1. Connect the drive to a running system. Read SMART values.

2. Do a SMART short self test. Do a SMART long self-test.

3. Zero fill / Wipe the drive with the manufacturer's utility. Entire drive. Preferably more than once.

4. Compare SMART values. If no anomalies, all good to go. Install your OS and continue. If problems, drop to DOS and do all the steps I outlined earlier just to make sure.

You can ignore hordes of anal-retentives who would talk about the time it takes to test a drive since you are just clicking few buttons and going on about your business in your existing system while the drive is being tested. You will save so much time (not to mention precious data [don't forget to do BACKUPs]) later on since it seems to me that a whole lotta defective drives are being shipped with transition to perpendicular recording.

Share this post


Link to post
Share on other sites

I've had a couple dozen drives the past 10 years. Didn't test any of them. Only the 2 DeathStar 75GXP's died on me.

So, taking into account that the 75GXP was a horrible series, it's best to NOT test your drives because... not a single of my untested drives one died on me.

Share this post


Link to post
Share on other sites

The whole point of burning-in the drive is to protect yourself from cases like the two that died on you.

I've had a couple dozen drives the past 10 years. Didn't test any of them. Only the 2 DeathStar 75GXP's died on me.

So, taking into account that the 75GXP was a horrible series, it's best to NOT test your drives because... not a single of my untested drives one died on me.

Share this post


Link to post
Share on other sites

Personally I don't like stress testing my hard drives. My procedure with a new drive is pretty simple and not time consuming. Here is what I do :

1. Run the drive and listen for any anomalies.

2. Quick format and install OS (or transfer data).

3. Check the SMART data for any anomalies.

So far I've not done anything that takes any time. This is because at this point I'm usually waiting on "it". As soon as possible (but when I wont be waiting around for it to finish, typically on the first night) I then :

4. Do a full surface scan eg "chkdsk x: /r"

5. Retest SMART data and compare with that of 3 to see if anything is deteriorating, especially looking for anything like re-allocated sectors.

After passing that the drive is in use but I always consider any new drive to be kind of "on probation" for the first month or so. Basically this just means I'm a little more cautious then usual about not having any non-backedup data on the drive and checking the SMARTs a little more often.

Share this post


Link to post
Share on other sites
The whole point of burning-in the drive is to protect yourself from cases like the two that died on you.

The problem with the 75GXP was not something that would have been unveiled by something like a stress test you do at home. I've had one (1) die in a customer's machine, an IBM Netvista something. Of ALL the other pc's of that series with 75 GXP's we sold - several hundred, not a single one died. Of all the 75GXP's in non-IBM pc's - 20-ish, at least half had to be replaced, and the RMA'ed drives as well. So an educated guess is that the problem with the 75GXP had to do something with drive firmware and system BIOS being incompatible. y

Share this post


Link to post
Share on other sites

IBM Deskstar's had mechanical problems and that is exactly the kind of problem you eliminate with a stress test.

I had one that died with a click of death during stressing and we sent several to customers and nothing happenned to them. So I was surprised when everyone was complaining all over the net. These issues do not affect anyone but the most ignorant. And unfortunately there are too many of you.

Share this post


Link to post
Share on other sites
The whole point of burning-in the drive is to protect yourself from cases like the two that died on you.
The problem is, of course, catching how much of the failures is likely? There's only so much testing you can do before you have to toss them into production.

How much is up to the system builder or to the user.

Share this post


Link to post
Share on other sites
IBM Deskstar's had mechanical problems and that is exactly the kind of problem you eliminate with a stress test.

Mechanical problems in non-IBM machines and virtually no problems in IBM machines? Now really...

These issues do not affect anyone but the most ignorant. And unfortunately there are too many of you.

Well thanks for accepting me into your club.

Share this post


Link to post
Share on other sites

I always exercise HDs before using them in a system. It is not a philosophical issue but a physical one. The surface of a platter has defects, and some of these will make sectors unusable. Most HD controllers these days are quite sophisticated and do Bad Block remapping internally. (I am so grateful I don't have to do this by-hand anymore.) Reading and Writing each sector on the whole surface of the disk a couple of times gives the controller a chance to encounter the blocks which are bad currently.

It is quite common for a disk to sit on a shelf for a year before being put into service. Some of the blocks which were only "weak" during the manufacturer's test will decay to "bad". I also suspect some disks never get a once-over at the factory. Especially "bulk disks."

These defects show up as "bit-rot" in the binary data stored on the disk. Changing a bit in the color representation of a pixel in streamed-video can be ignored. Changing a bit in an executable program of the OS cannot. As disk sizes have increased, the fraction used by the OS has gone down, so the probability of one hitting the OS is smaller. However, areal size of each sector has also gone down, which increases the probability that a defect will be "big enough" to limit the usability of a bit.

The Heads never touch the surface of a disk. At least, not in a working disk. You do not "wear" a disk by using it. You do "wear out" the weak areas by flipping the magnetic domains, so it is best to do that before you plant data on it that you want to read back later.

So, always exercise the partition on which you will install the OS.

Share this post


Link to post
Share on other sites
It is quite common for a disk to sit on a shelf for a year before being put into service
I assume you mean in your specific use? Just clarifying.
These defects show up as "bit-rot" in the binary data stored on the disk.
True, non-recoverable bit error rates are usually 1*10^15 reads or better, so the odds of it being a problem are pretty low in most single-disk configurations.
ou do not "wear" a disk by using it. You do "wear out" the weak areas by flipping the magnetic domains,
Your first statement is a little misleading... just keeping things clear.

Share this post


Link to post
Share on other sites

When you receive a new drive, simply do a quick manufacturers test to make sure it survived shipping.. The words "burn in" do not apply to hard disk drives, nor do terms like "break in". There's only "test" and "exhaustive test".

Receive drive, test drive, put it to use.

F

Share this post


Link to post
Share on other sites
When you receive a new drive, simply do a quick manufacturers test to make sure it survived shipping.. The words "burn in" do not apply to hard disk drives, nor do terms like "break in". There's only "test" and "exhaustive test".

Receive drive, test drive, put it to use.

F

I have to agree with you on this. I have never done a "burn in" test on any drive I have installed. It either works or it doesn't. I generally do a initial diagnostic on the drive before I use it and if it passes, then I put it into service. I think if it makes you feel better you can do all these tests. All you are really doing is putting unnecessary wear on the drive. My rule of thumb on hard drives is this, it either works or it doesnt. Does not matter what brand it is. So doing a burn in test doesn't make much of a difference. Just run some tests and be done with it.

Share this post


Link to post
Share on other sites
When you receive a new drive, simply do a quick manufacturers test to make sure it survived shipping.. The words "burn in" do not apply to hard disk drives, nor do terms like "break in". There's only "test" and "exhaustive test".

Receive drive, test drive, put it to use.

F

I have to agree with you on this. I have never done a "burn in" test on any drive I have installed. It either works or it doesn't. I generally do a initial diagnostic on the drive before I use it and if it passes, then I put it into service. I think if it makes you feel better you can do all these tests. All you are really doing is putting unnecessary wear on the drive. My rule of thumb on hard drives is this, it either works or it doesnt. Does not matter what brand it is. So doing a burn in test doesn't make much of a difference. Just run some tests and be done with it.

Share this post


Link to post
Share on other sites
IBM Deskstar's had mechanical problems and that is exactly the kind of problem you eliminate with a stress test.

It wasn't a mechanical problem, it was IBM's "pixie dust" treatment of the disks that caused clogging of the heads, which in turn caused overheating when the heads idled over the same tracks, tracking issues and data loss.

Your tests are still pointless, all you're doing is waving that frog around and any evidence you've presented is anecdotal. "I did this to my drives and none of my drives failed (except the ones that failed anyway.......)" Post hoc, ergo propter hoc logic fallacy; you've no idea if those drives would have worked just fine anyway without going through your motions! :P

Any stress testing your tests do is not going to scare away any possible faults that might affect that particular unit. A stress test won't eliminate forever the chance of a potential head crash. Essentially, the task of installing windows vista and patching it up with windows update to SP1 and current security fixes, installing some autostarting programs, and then (re)booting the machine a couple times is going to do all the stressing you'll ever need of that drive.

...And that WITHOUT the needless wear and tear you'd introduce by jumping through all the hoops you've described so far...

If you're worried about your drive crashing on you, what you need is not waving frogs in the air, but solid backup procedures and a redundant disk array.

Share this post


Link to post
Share on other sites
I disagree with both 6_6_6 and FaaR, partially only.

I've been building machines for 18 years now and I can assure you, I knew 6_6_6 was on the money the second I read his first post, purely from the wording of it.

This guy clearly knows what the hell he's talking about.

Yes, drives can still fail after a stress test but what he's doing is thoroughly thrashing the drive, if it works after all that (and hopefully he keeps it actively cooled) then it might reduce 3 months from the overall life of the drive BUt it's very likely to weed out a lemon, which can be all important sometimes.

I wish I had done what he does more often (and thanks to his post, I'm doing some of it right now, these Seagate 1.5TB's need it before you can trust them)

Share this post


Link to post
Share on other sites
Wow, seven posts of useless drivel that is not only utter rubbish but of no relevance at all. Well done - glad you had nothing better to do for four hours. The lack of maturity you demonstrate here and in the other thread where you've resorted to personal attacks shows that nothing you say can be taken seriously.x

I might be 3 months late to this thread but it looks to me here there's one person throwing around rubbish posts and that would be the person accusing others of 'useless drivel'

Just because a good thrashing won't weed out all bad disks, doesn't mean it might not save your butt on a drive which is a bit ho hum.

Being sure the drive can be worked, before you deploy it could save pain later.

Share this post


Link to post
Share on other sites

How anyone in this thread can be debating with this 6_6_6 guy I don't know.

I've been working on machines since I was 13 years old (ok PC's at 13, computers in general at 10 y/o)

I'm 31 now, I've genuinely had at least 100 hard disks of my own in the past 18 years, of those 100 drives only about 5 or 6 have died (and I do NOT stress test)

Now,... 2 of the failures were IBM deathstars

1 was a Maxtor

2 were WD drives, died within 2 days of ownership (JUST after I'd finished moving all my data on to them >:( )

1 (right now) appears to be one of the shady, Seagate 1.5TB jobbies.

So the figures state, that overall Iv'e saved a hell of a lot of time by not thrashing my disks, since mosto f the time they didn't die.

None the less I HAVE lost data and it pissed me off no end.

Now if I could have isolated anyone of the drives before it died on me and not put data on it in the first place, that would be a good thing.

You guys are arguing with this dude, when infact now more than ever we should be testing these bloody things, the Seagate 7200.11's have a terrible name for failures, how hard is it to thrash a drive around for 36 hours before putting data on it?

Sure if it's for a customer or a friend or family member or work, don't bother but your own, important data? cmon!

Yes RAID can solve this, yes backusp can solve this, but backing up 1.5TB isn't a trivial operation.

In the past 5 years I've had the following system

200gb HDD x 3 sold then got ...

400gb HDD x 3 sold then got ...

500gb HDD x 3 sold then got ...

750gb HDD x 3 sold then got ...

1.5TB HDD x 3

etc..

This time, I intend to properly thrash the heck out of these 1.5TB drives before I throw data at them, they are passively cooled and on a high quality power supply

One drive has ALREADY started clicking and they've been in the machine 12 hours .... now it didn't start clicking till it was 3 hours old so if I was dumb enough to move data on it already - well

How hard is it to use a spare machine (or alt tab) and leave some operation in the background checking your drive?

Seriously, sure his information won't find EVERY FAULT EVER but it's going to isolate a hell of a lot of them, it's simple logic - there's a reason proper PC stores used to do this in the 70's, 80's and early 90's

Share this post


Link to post
Share on other sites

abrasion, thanks.

It was not only being used in 90's... it is still being used by manufacturers on Enterprise drives part of their Quality Control process.

Today I just returned a 1TB Seagate I bought for myself. It was fine for random writes, handled tests fine for couple of hours. I did an entire surface erase and when I came home the drive was no longer online. Few weeks ago a collegue brought her 640GB WD... she was complaining random vista crashes. Drive was toast after a night of torture.

I have not lost a bit of data in my entire life. And I am not planning to. I would think twenty times before committing any data on any of these high-density drives. I have no desire to waste no more time on hordes of ignorant rednecks. Well, each to his own.

Edited by 6_6_6

Share this post


Link to post
Share on other sites
I have no desire to waste no more time on hordes of ignorant rednecks.

Somehow, I don't think 6_6_6 will ever win Miss Congeniality, but I do appreciate the technical info he/she brings. Thanks.

P.S. Here in Virginia, the term redneck can be taken as a compliment. Thanks again. :)

Share this post


Link to post
Share on other sites

I am sorry, I did not want to offend real rednecks. I am not from USA and I was refering collectively to redneck posers/pretenders/imitations. I also did not mean to use 'redneck' as a derogatory term (Isn't that 'hickey'?), instead wanted to emphasize on 'ignorant'.

P.S. Here in Virginia, the term redneck can be taken as a compliment. Thanks again. :)

Share this post


Link to post
Share on other sites
I am sorry, I did not want to offend real rednecks.

Hmmm... calling someone a "real redneck" might not be the best way to apologize for calling them a redneck, IMHO.

No worries, 6_6_6, I wasn't offended at all by your comment. You always seem to liven things up a bit here.

P.S. A "hickey" is something else entirely (see dictionary.com - definition 1a is wrong, 1b is correct). Perhaps you meant "hick." Personally, I'm not sure which word would be worse (for your health/safety) if you called someone a "hick" or a "redneck" or "ignorant", be it true or false, in southern or western Virginia....

Share this post


Link to post
Share on other sites

Well either way thanks to your suggestions I'm going to thrash the jeebies out of these disks.

I'm gonna try IOMETER I think to do some thrashing but it's taken near 5 hours just to write the file to the drive to do the thrashing!

Share this post


Link to post
Share on other sites

Yes, you are right. 'hick'. Well, i am a bit of redneck myself, love country music and all. Emphasis being on 'ignorant', i am sure my point is understood.

I am sorry, I did not want to offend real rednecks.

Hmmm... calling someone a "real redneck" might not be the best way to apologize for calling them a redneck, IMHO.

No worries, 6_6_6, I wasn't offended at all by your comment. You always seem to liven things up a bit here.

P.S. A "hickey" is something else entirely (see dictionary.com - definition 1a is wrong, 1b is correct). Perhaps you meant "hick." Personally, I'm not sure which word would be worse (for your health/safety) if you called someone a "hick" or a "redneck" or "ignorant", be it true or false, in southern or western Virginia....

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now