Best way to test refurb HDDs?

[VirtualLarry]

Just wondering what other people use, when they buy refurb HDDs.

I usually use WD's Data Lifeguard for DOS, and run the Short and Long self-test, and then slave them to a Windows 7 machine, and do an NTFS full format.

This time, I don't have Windows on my testing rig, so I'm booting off of a Linux Mint 17.1 disc, using "Disks" tool, and doing the Short and Extended SMART self-tests.

 

[BonzaiDuck]

Just wondering what other people use, when they buy refurb HDDs.

I usually use WD's Data Lifeguard for DOS, and run the Short and Long self-test, and then slave them to a Windows 7 machine, and do an NTFS full format.

This time, I don't have Windows on my testing rig, so I'm booting off of a Linux Mint 17.1 disc, using "Disks" tool, and doing the Short and Extended SMART self-tests.

I don't buy refurbs, but there's always a first time for anything. With HDDs, it just doesn't seem to make sense, unless you get a slew of them for a ridiculous price-per-unit..

I use the WD Diag/ or data Lifeguard for their disks. For Seagate, Seatools. They all offer their own diagnostic program. Mostly I recycle my own old disks into new uses. And I have a cache of them in my parts-lockers -- more in USB/eSATA boxes and swap-caddies.

But like we may have observed in an old exchange, "I either have too much 'stuff' -- or a lot of extra 'stuff' in excess of what I can use at a given moment.

 

[Quad5Ny]

Just wondering what other people use, when they buy refurb HDDs.

I usually use WD's Data Lifeguard for DOS, and run the Short and Long self-test, and then slave them to a Windows 7 machine, and do an NTFS full format.

This time, I don't have Windows on my testing rig, so I'm booting off of a Linux Mint 17.1 disc, using "Disks" tool, and doing the Short and Extended SMART self-tests.

Extended SMART test is good but I think it only try's to read each sector, not write.

After that's done you can use "clean all" in Windows diskpart or a full format (not quick) in your favorite Linux distro to do a write test (maybe follow that with another read test ).

 

[dave_the_nerd]

dd if=/dev/zero of=/dev/[hdd identified]

If it crashes or breaks instead of completing, the drive is bad.

 

[Quad5Ny]

dd if=/dev/zero of=/dev/[hdd identified]

If it crashes or breaks instead of completing, the drive is bad.

That's only a write test. You would have to do "dd if=/dev/sd_ of=/dev/null" afterwards.

Its also be a good idea to set the block size to something large like 8MB to speed up the test as dd's default is 512bytes.

 

[Elixer]

I like doing a full format (with dd or whatever) then use bonnie++ for a few hours.

 

[VirtualLarry]

I don't buy refurbs, but there's always a first time for anything. With HDDs, it just doesn't seem to make sense, unless you get a slew of them for a ridiculous price-per-unit..

$13.99 ea, got four of them. 160GB WD SATA2, 2MB cache. Not speedy.

So far, out of three I've tested, all three have been good. (No re-alloc. sectors.)
Two HDDs had 5 years worth of POH. One had 3 years. All of them had around 60 power-cycles. Likely some kind of server drives, which would be a good thing.

Edit: All four tested out good. Three out of four had 5+ year POH times. One had 3 year.

 

[BonzaiDuck]

$13.99 ea, got four of them. 160GB WD SATA2, 2MB cache. Not speedy.

So far, out of three I've tested, all three have been good. (No re-alloc. sectors.)
Two HDDs had 5 years worth of POH. One had 3 years. All of them had around 60 power-cycles. Likely some kind of server drives, which would be a good thing.

That seems like a "ridiculous price per unit" OK. Not sure what you'll use those for. Almost all the SATA-2 drives I have now are 500GB. I have some SATA-3's that are 320.

 

[VirtualLarry]

Boot disks for Linux Mint 17.1 machines, to give away.

 

[unixwizzard]

$13.99 ea, got four of them. 160GB WD SATA2, 2MB cache. Not speedy.

So far, out of three I've tested, all three have been good. (No re-alloc. sectors.)
Two HDDs had 5 years worth of POH. One had 3 years. All of them had around 60 power-cycles. Likely some kind of server drives, which would be a good thing.

That doesn't sound like they were refurbs, more like simple pulls.

I've received refurbished drives from Seagate and WD as RMA replacements, and all of those reported only around an hour or two of run time.

I certainly would not trust an unknown drive with 5 years of POH..

 

[Red Squirrel]

dd if=/dev/zero of=/dev/[hdd identified]

If it crashes or breaks instead of completing, the drive is bad.

This is what I do too, on top of running smart tests. I do a read test and a write test.

So basically something like this: (we'll say the drive is sdb)

smartctl -a /dev/sdb

Check existing smart stats, if there's already errors then it's a dud and don't waste your time.

dmesg -c

Just clear the dmesg log. (easier to check it later)

Then write test:

dd if=/dev/zero of=/dev/sdb &

Then every now and then type dmesg to see if there's any weird I/O errors.

Once all is good then do read test:

dd if=/dev/sdb of=/dev/null &

Then every now and then type dmesg to see if there's any weird I/O errors. Then once this is done check smartctl -a again to see if there's new errors.

After this it's probably safe to say the drive is good. Would be a good time to run the smart tests if you want.

 

[bryanl]

MHDD. It shows slow sectors, not just totally failed ones.