SPCR

so im curious about what people do to make sure that their newly purchased hard drive is not a partial lemon before entrusting it and placing a ton of data on it.

the reason i ask is, i just bought two hard drives to replace my existing ones.
what would i need to do such that i can trust it not to corrupt my data or die.

one idea is to backup to another drive,
but what if this new drive is suppose to be the backup drive,
what do you do in that case?

do people run the diagnostic tools from the hard drive manufacturer?

You can do a full format to check for bad sectors and manufacturer diagnostics. Neither will guarantee you that the drive is safe to use though as they do sometimes fail without any symptoms.

Have a backup and when you lose if it or the source drive dies or malfunctions, get another backup in place ASAP.

Download HDDScan. Run "extended SMART test", then check for SMART values.

You can do full format but extended SMART test does the same (and a bit more). Manufacturer diagnostics isn't necessary as if you run manufacturer's diagnostics (which usually requires you to boot to DOS) all the utility does is give the HDD an order to perform autonomous self-test of it's own condition. Why not do it in Windows as you can continue to use the computer normally while the HDD is being tested? The test itself is identical regardless whether you use manufacturers utility to initiate self-test or whether the self-test command is used from within your regular OS. Latter one is just easier and doesn't prevent you from productive work while it runs quietly on the background.

At the moment, I happen to be running one of my periodic SMART self-tests for all my HDDs at once. 6 HDD doing extended self-test. Should be finished by tomorrow morning. No manufacturer utility allows simultaneous testing of multiple drives let alone testing drives of other manufacturers'. HDDScan does that.

Yeah, and it's a freeware, of course.

This far I have done nothing at all, except not remove the content of the old drive when I clone it to a new drive. Then, if the new drive fails, I still have the old one. After some time has passed, I sell the old drive or put it to some other use.

thanks for the advice

i dont mind booting into dos as im accustom to it and making a usb key boot into dos isnt hard at all
and the machine that these drives are going into doesnt have any space left to mount extra drives, though i do have another i plan on using to run these tests

i too never have done anything special before
it is just that i have more data and maybe a bit more paranoid

whiic wrote:Download HDDScan. Run "extended SMART test", then check for SMART values.

You can do full format but extended SMART test does the same (and a bit more). Manufacturer diagnostics isn't necessary as if you run manufacturer's diagnostics (which usually requires you to boot to DOS) all the utility does is give the HDD an order to perform autonomous self-test of it's own condition. Why not do it in Windows as you can continue to use the computer normally while the HDD is being tested? The test itself is identical regardless whether you use manufacturers utility to initiate self-test or whether the self-test command is used from within your regular OS. Latter one is just easier and doesn't prevent you from productive work while it runs quietly on the background.

At the moment, I happen to be running one of my periodic SMART self-tests for all my HDDs at once. 6 HDD doing extended self-test. Should be finished by tomorrow morning. No manufacturer utility allows simultaneous testing of multiple drives let alone testing drives of other manufacturers'. HDDScan does that.

Yeah, and it's a freeware, of course.

from your experience, how long does an extended self-test run take?

If it is a rotating disk not an SSD I'd say HDtune error scan (not the quick version) then put the drive into real use keeping the old drive in reserve in case the new drive dies. Watch the smart data using HDtune or Speedfan on a regular basis for the first few weeks.

Save any more thorough diagnostics for after you find a problem using the convenient general purpose tools.

"from your experience, how long does an extended self-test run take?"

I don't remember exactly. It takes the same amount of time as verify pass (+ one or two minutes to do mechanical testing) because that's basically what extended self-test is.

For 1TB GP (1st gen) it takes roughly 5 hours to self-test. For 2TB GP (3rd gen) it takes roughly 7 hours to self-test. Self-testing all my HDDs would take around: 5+5+4+4+4+7 =29 hours. For this obvious reason, I do not run manufacturer's diagnostics in DOS. I prefer running all the self-tests simultaneously while booted to my OS and instead of 29 hours, it will all be over in 7 (some drives faster than that). And computer remains usable while it's doing it (but usually I do set it to self-test overnight to avoid performance loss).

i followed your advice and i tried by hdtune and hddscan
they both took about the same amount of time

4hrs for 1.5TB drive
and 6 hrs for 2.0TB drive

the weird thing is that when i ran it on one of the empty 2.0TB drive, it passd the test, but as soon as i tried writing data to it, it started to exhibit issues of bad sectors and another scan indicated it was the case

likely the other 3 drives were fine, both before and after writing data to it

If you really want to test a HDD before use it will take some time.

Give the drive a quick check.
Give the drive a full check.
Write One's to the whole drive.
Give the drive a full check.
Write Zero's to the whole drive.
Give the drive a full check.

All of the checks are there to ensure that a fault is found after a change in the magnetic state of the drive's surface, and before the state changes back to where a fault might possibly not be diagnosed (it does happen.

A quicker option would be to write One's to the entire drive, test, then use if OK. Remember that drives leave the factory covered in Zero's, so you will change the state of every bit on the drive before testing, but you have not then changed them back, so this is only really a half cycle, but far better than just a test alone.

This is what a neurotic colleague of mine does, and sinse he started doing the above full test cycle, he has not had a drive fail on him in use, they have all failed during testing - he has a silly amount of drives. It is worthwhile if you can spare the time, I am part way through the process on 2x 2TB drives, all good so far.


Andy

is there a program that does this?
or are you using a script in linux/freebsd/...

Instead of doing extensive initial testing, I tend to fill the drive up with data right away... but NOT deleting the originals. Then I keep the new drive as an online back-up and run extrended SMART test weekly (or more often) for a period of month or two. If the drive is still in perfect condition, I do a checksum test to check that there's no silent data corruption that has gone unnoticed during SMART test. After it I consider the drive passed and might delete originals (if they are not important to keep a back-up even after the new drive has been thoroughly tested).

Elevated failure rate last for a couple of weeks/months after taking the drive into use. Don't believe andyb blindly in the claim that running extensive test for day or two is enough. Naturalle overwriting it a couple of times give it more (or lest say "different kind of") stress than just reading it (which is what SMART self-test does).

Lately I've been testing drives with a couple of programs from the Ultimate Boot CD. I start off by running the MHDD32 seek test for an hour, though it seems to have a bug that causes it to not work on 1TB and larger drives. Then I use Darik's Boot and Nuke do multiple write/read tests. I set the wipe method to pseudo-random and tell it to verify the data on every pass. I highly recommend doing this on a spare machine with no other drives installed to avoid overwriting any important data. Also, make sure that there is some airflow over the drive because this testing is pretty intensive. It also takes a while ...

whiic's advice about retaining the old drive and comparing checksums is also good, though I suspect that read errors are a lot more likely than checksum mismatches due to silent data corruption. In either case, you can still restore the data from the old drive if you still have it. The next step past this is RAID.

I now think that running an extended SMART test periodically is a good idea. I recently replaced a 2 1/2 year old Samsung drive that silently developed a number of unreadable sectors in a handful of files that hadn't been accessed in the past year or so. A periodic SMART test would have found the problem sooner, and might have reduced the amount of data lost.

is there a program that does this?
or are you using a script in linux/freebsd/...

I use the Seagate test tool (SeaTools for Windows - Long Drive Self Test) for the testing - the Zero' and One Writing is done by a piece of software a colleauge wrote for the purpose, its a bit clunky, but it works.

Both of these can be run on several drives in parallel, but your southbridge might cause it to run very slowly, others will run at/near full speed with several drives.


Andy