How do you people test brand new hard drives?

[Doomer]

OK, this is bordering on the ridiculous. a significant percentage of the new drives I'm getting in these days are defective. My most recent order was for 5 enterprise class WDs. 2 were defective and I had to send them back. four of them went into a Synology NAS. I was in a hurry so I didn't test them first. The NAS told me one was defective. The other wasn't detected by the computer. These are not consumer grade HDs, these are supposed to be better grade HD. But 2 out of 5 bad ain't better. For Christ's sake, these drives are 300 bucks each. One should expect better.

So how do ya'll test a brand new out of the box HD to ensure it's healthy. I normally do a full format and then run scandisk. Is there a more accurate way to determine that a HD is good for the long term?

 

[Ketchup]

I few years ago I had a guy tell me drives don't last like they used to, and I do not doubt it. I think the one Bubbaleone mentioned in this thread might serve you well:

http://forums.anandtech.com/showthread.php?t=2336581

Passing the extended test and the SMART test should be a pretty good sign for the drive.

 

[Elixer]

You do a intensive stress test, like bonnie++ (linux), or something that will read/write to every sector of the HD a few times... then, in theory, it should last longer than something that wasn't tested fully. One pass full format, while better than nothing, doesn't seem to weed out the dud HDs as well as doing stress tests on them.

Problem is, the quality has sunk to a all time low IMO, and the only way to fight this is keep backups. Yeah, I know that isn't what your asking, but, these new drives fail much quicker than HDs of the past.
I still think that packaging of the drives play a big role in this as well.

 

[corkyg]

As far as I can remember, I have never tested a new drive unless it exhibited some reason to do so. That goes back to 1992.

 

[Tsavo]

As far as I can remember, I have never tested a new drive unless it exhibited some reason to do so. That goes back to 1992.

I give all my drives the Voight-Kampf test.

So far, so good.

 

[smakme7757]

No point wasting time testing drives. I just put them in and use them. If they go bad i replace them and restore from backup.

 

[hot120]

Use the Data Lifeguard tool set from Western Digital.

 

[Zap]

As far as I can remember, I have never tested a new drive unless it exhibited some reason to do so.

:thumbsup:

HDDs will usually exhibit very obvious problems. Seems as if a HDD that can detect without issues and can accept a quick format usually does not have any problems that immediate testing will expose.

 

[iluvdeal]

I do a full format then thorough scan with the manufacturer diagnostic.

 

[WilliamM2]

I use IBM/hitachi Drive Fitness test, as it works on all brands. I run the full test, which can take 3 hours or more on larger drives.

I have had brand new drives which appeared to work fine, but the test showed bad sectors. They get returned immediately. I've only had 1 or 2 drives that passed the test fail at a later date, and that's out of hundreds of drives.

 

[_Rick_]

Either I write it full with random data, or I rebuilt a RAID onto it.
Then I can check write speed, and have a quick look at reallocated sectors, if I'm feeling anal. Most of the time, I'm happy if the rebuild / dd just runs through, and the next surface scan (I think I run one on my RAIDs on a weekly basis, to make sure read errors don't accumulate) doesn't throw it out.

So basically, I don't test them, and I don't have to, because I don't really trust them.

 

[Ketchup]

I use IBM/hitachi Drive Fitness test, as it works on all brands. I run the full test, which can take 3 hours or more on larger drives.

I have had brand new drives which appeared to work fine, but the test showed bad sectors. They get returned immediately. I've only had 1 or 2 drives that passed the test fail at a later date, and that's out of hundreds of drives.

This is exactly what I was thinking of with the OP. As long as Windows can format the drive, there could be a bad sector and you would never know until your data reaches it. When you are using drives like you and the OP are describing, I think a test like this would be well worth the time spent. Especially if you put a bunch in a NAS and take care of them at one time.

 

[WilliamM2]

This is exactly what I was thinking of with the OP. As long as Windows can format the drive, there could be a bad sector and you would never know until your data reaches it. When you are using drives like you and the OP are describing, I think a test like this would be well worth the time spent. Especially if you put a bunch in a NAS and take care of them at one time.

I didn't even know that a Windows version existed, looks nice. I've been using the DOS/Linux version that comes on the Ultimate Boot CD for years.

 

[jjsbasmt]

HD Tune Pro my program of choice especially the excellent SMART checks and the full drive sector scan. I check EVERY HD new or used on every system I work on.

 

[Magic Carpet]

I use IBM/hitachi Drive Fitness test, as it works on all brands.

What version, care to remind? The ones widely available (3.77 - 4.16, minus WinDFT) on the net, only work with Hitachi branded devices.

 

[corkyg]

What version, care to remind? The ones widely available (3.77 - 4.16, minus WinDFT) on the net, only work with Hitachi branded devices.

I just used IBM DFT 4.08 (also Hitachi) on a bootable floppy. I have no Hitachi frives, and it ran just fine and could do my Seagate OS drive with Win 8.

 

[thecoolnessrune]

I used to use bonnie, but now I have my array all set up, so for zpool expansions or replacements, I just put in another hard drive and let it go. If its a replacement, then the workout from rebuild will give it a good testing. In both cases, a failed drive isn't a big deal, and can be dealt with if/when the zpool says its degraded.

 

[capeconsultant]

My first test is... "Does it have SSD in the name?" I know, that does not guarantee much, but I thought it was funny

 

[philipma1957]

I PUT THEM in my pegasus r6 in a raid0 then I use mac osx disc utility to overwrite it with 0's 2x 2th = a 4th raid0 . do 7 x over write slow since it is about 27th worth of 0's at about 300 , but if they pass this they do not go bad.

 

[Red Squirrel]

I noticed too that drives are really not as reliable as before. When you get a good one, it's a good one, and it will last, but often, they are DOA or die very fast. Always order 30% more drives than you need. If they actually happen to all be good or after you RMAed the bad ones, then you have spares.

I don't know if this is the best test or how effective it is, but what I do is run a dd on it to read every single sector. Then I do another dd to write every single sector. I just zero it out. I keep an eye on dmesg for anything weird, and I also look at the smart data for any error counters. Often a drive will pass the smart test but still be failing/failed. Doing a smart test is still a good idea though but can't be depended on 100%.

 

[dighn]

what do you mean "you people"?! i resemble that.

I used to run a full scan every time I used a new drive for something, but with the capacity of drive these days, that takes forever, so I don't do it anymore. I'm just about to pull the trigger on a new virtualized home server, but I'll be using ZFS so I'm not too worried.

 

[ch33zw1z]

If I do any type of testing on new drives, it's to check the SMART status and a full format. If it is acting weird, running manufacturer diagnostics is the next step.

 

[flexy]

I haven't sector by sector "tested" a HD in many years. The fallacy here is that a test on a NEW drive cannot predict that your HD may not fail next week or next month. A xyz hours lasting test doesn't guarantee you that the drive may not fail next week. It doesn't guarantee that something might not happen like an unforeseen power outage etc. or whatever else which MAY result in a bad sector..AT SOME, RANDOM time you cannot predict.

HDs, especially SSDs have a tendency to fail "in time", they may get worse gradually. For that reason I think it's really better to have regular backups in place (or if you're really paranoid use a tool like HDSentinel and monitor your SMART stats, it's pretty good)..ALTHOUGH I can see *why* some people would "test" their drives before they put them in systems. I just personally don't have the nerves/patience. I do testing THEN when an obvious problem occurs (of course assuming I have a recent backup always stored) - and THEN do a "disk re-initialization" test with HDSentinel to make sure that the bad sector is found and reallocated.