Re-using old drives for storage

[silverwolf0]

So I have some older hard drives I want to wipe, test for any errors, and properly format to be used for basic storage. Can I get a quick rundown of what all I should be doing? I want to make sure the drives are in good working order.

 

[7x1x]

Check the SMART data, look for bad sectors among other things, and run some self-tests. Make sure you have two copies of any important data on separate devices.

 

[PliotronX]

Record the smart data as 7x1x suggested and perform full formats (not quick). I've had full formats mark pending sectors as good but this will also allow them to be reallocated if tested bad by the drives logic. Beyond that, if no odd mechanical sounds are emitted they should be G2G!

 

[sinisterDei]

I have to do this occasionally for work- we generate a bunch of hard drives we send back and forth in the mail and eventually they end up in a vague pile and nobody's really sure if they're good or not.

Check the SMART, make sure they don't make any noises, but if you've got the time and available SATA ports I run them through DTI's Windows Surface Scanner tool. It's free, and it checks every sector. Helpful hint, you can connect a bunch of drives to a system at once and open up multiple copies of the surface scanner so you can run them all at the same time. I've got an old Athlon X4 workstation set up as a hardware test bench with sixteen SATA connections (and a shit-ton of SATA power Y-cables) coming from it so I can plug in as many drives as I have on me at a time to test. It takes a good while to run each drive, but then I just write "Known good 12/2017" in sharpie on each drive when they check out, or toss the failures in the bin if they're out of warranty.

 

[SamirD]

A well-built drive will last well beyond its warranty, but the question is not if it will fail, but when. So if they're used for general storage, I primarily check the date of the drive and warranty as they are now being manufactured for planned obsolescence. And then I plan accordingly for the failure event--does that mean raid1 or mirroring, or selling them off and using the funds for new drives, etc.

A good thorough scrub and format will find any issues that are present at the time, but it can't help predict anything down the road (except that a drive with bad sectors has an increased chance to have more). Regularly testing them requires some time, but I think a failure plan is more effective because ultimately if you know a drive is going to fail, you're going to be enacting a failure plan anyways--might as well just plan for that, format the drive and hope for the best.

 

[sinisterDei]

a drive with bad sectors has an increased chance to have more

Even one bad sector a utility manages to find (not a SMART counter, but an actual bad sector a surface scanner encounters) is grounds for tossing it in the bin to me.

 

[SamirD]

Even one bad sector a utility manages to find (not a SMART counter, but an actual bad sector a surface scanner encounters) is grounds for tossing it in the bin to me.

That's sad because in many applications a drive like that will be perfectly acceptable. I'd take any drive like that as I have a lot of applications where a drive like that will work fine for years.

It's also important to note that there are many, many spare sectors that bad sectors can be swapped with, and sometimes this is reported to the OS and applications and other times it is not.

 

[sinisterDei]

Yes, but if you can detect the bad sector in a third party utility that isn't just checking a smart counter, you've hit your limit

 

[Brian_B]

Yes, but if you can detect the bad sector in a third party utility that isn't just checking a smart counter, you've hit your limit

With traditional HDDs, my experience has been as soon as you get that first bad sector, it starts to snowball quickly. Not always, but often enough to make me agree with sinisterDei and just chuck it. Or if it makes the "Washing Machine Sound of Doom", your done.

SSDs, I haven't had so much trouble with. They are designed to fail sectors gracefully (overprovisioning), so it's just part of their lifespan. With SSDs, it seems they keep chugging until the controller just up and dies (and your screwed).

 

[sinisterDei]

I mean, HDDs are designed to gracefully fail sectors. But that's the key- gracefully. The only way you can even tell they did it is by checking SMART counters. If you run a surface scanner, it'll just glide right over the bad sectors because they've been remapped. If you run a surface scanner and it *finds* a bad sector, that's because it *hasn't* been gracefully remapped, and it's time to ditch the drive if you care about your shit even a tiny bit.

 

[SamirD]

I understand what you guys are saying, but for example, I have one drive in operation that past 10GB (total 20gb) it's nothing but bad sectors. I've been using it as a 8gb drive for a few years now with no problems. It's not the typical application, but they are out there.

Anyone chucking drives like this, just send them my way. I'll let them live on for a while doing some good before they give up the ghost.

 

[Brian_B]

Depends on your willingness to roll the dice. That drive your talking about could run another 10 years no issue. It could die tomorrow.

With proper backups, it's just a matter of nuisance. How much that's worth to you depends.

 

[SamirD]

With proper backups, it's just a matter of nuisance. How much that's worth to you depends.

Exactly, and that's the way all drive failures should be--just a nuisance.

I'll typically use a bad sectored drive as boot drive for a kiosk application where most of the time the drive is not written to since the kiosk writes to memory and then removes it upon reboot. Drives seem to work fine for years in this config, and when they inevitably fail, you just re-image a new drive and swap in the 'new' one. Easy-peasy.