3TB Seagate Barracuda 7200.14 ST3000DM001

[Yw-slayer]

PSA, mine just died, went online to Google failure just for kicks, and saw this: https://www.backblaze.com/blog/best-hard-drive/

Change yours now. I've got 5, but fortunately(?) 4 are in a Windows RAID array, so I can just stick in a 4TB drive in case one of the 3TB one fails, then replace them as they die. At least in theory.

[thesameguy]

I have a mountain of these drives in service, probably 30. But as a matter of course I never run an ATA drive more than three years, and endeavor to replace them every two years. AFAIK, no consumer drive can be trusted more than two years. After 24 months, start making a replacement plan.

[Yw-slayer]

How do you do it? All on RAID and then just pull them when they fail and replace with another one?

[Leon]

My Seagate external just shit itself.

[Yw-slayer]

Looks like the Deskstars are the most reliable now. A bit of a change from the days when they were called Deathstars.

[mk]

Those STx000DM001 ones have bad transient diodes.
Remove them.

You can strip drives from externals cases.

Those diodes are for over voltage surge and have a tendency to short.
If that is the case and you connect the drive to pc-power its circuit board will burn for sure.
1062

[Yw-slayer]

Wait, they'll damage the MB too?

[thesameguy]

No, they just fail and then the drive doesn't work anymore.


How do you do it? All on RAID and then just pull them when they fail and replace with another one?

Depends on how they are set up. For a RAID I'll just replace one member at a time until they're all replaced. Takes a long time that way, but it requires basically no intervention. Sometimes they're just lone drives in which case if they're live they'll be imaged to a new drive or if they're backup drives they get taken out of rotation.

I've watched enough hard drives tank over the years that I really didn't need backblaze to confirm what I already knew - ATA drives are just not built to last. Worse, it's not like there is a single manufacturer that always makes bulletproof drives so every purchase is pretty much a gamble so I have no brand loyalty. Much like backblaze, I buy whatever is cheap at the time because I know it doesn't need to last. Every couple years I got a storage bump and a relatively low cost and I don't have to deal with failures.

I think I mentioned it here somewhere, but around the year two mark I get a new drive and throw the old up on CL or ebay or whatever and offset the cost of the replacement. I'm sure in the end I spend more money at home on drives than I probably have to, but somewhere around $200 ($300 on drives, $100 back from sales) every two years for security isn't worth getting my feathers ruffled. I just actually replaced the three Seagate 3tb drives in my desktop with 4tb HGSTs a couple months ago. I wish I'd waited, as 5tb Seagates are cheap right now. Oh well!

In the office we have a mountain (like, 20tb) of data that is used by only a few people and I can't justify buying very expensive SAS drives - a 1.8tb SAS is like a grand. Although our external consultant gets all prickly about it, the whole shebang is stored on a couple RAIDs built of, right now, Seagate 3tb ATA drives. I've had one failure in the last two years and I'm approaching the time they'll be replaced. I'm really torn between those HGST 4tb drives or the Seagate 5tb drives. 4tb drives would need a 1-for-1 replacement, but the 5tb drives could potentially reduce the number of drives in play. I like efficiency, but more drives do provide some additional data security. We'll see where things are in a month or two.

Takeaway:

1. Don't trust any particular brand or model of drive. Everyone fails.
2. Toss ATA drives somewhere between the 24 and 36 month mark.

:)

[mk]

Wait, they'll damage the MB too?
No, only the circuit board of the *drive* will burn.

The catch is that the interface of the drive is "dynamic."
You can't swap them, period.
And if the content, that is still completely ok, includes the only ones of the world...

So, if the case is that the external drive has just shorted diodes you can still save the content by just removing those diodes.

[Yw-slayer]

Right. It's not an external drive, though.

Anyway, I can see the value of changing drives every 3 years. Storage gets cheaper all the time anyway.

[mk]

External came from Leon.

Key here is current limiting.
If not done then shorted component will grill the board.

Burned board can be replaced of course but it needs some settings from the old one.

Generally my take is that all moving while used stuff with mechanical drive should be treated like socks.
And backup accordingly.

[Yw-slayer]

I bought a sweet 6TB HGST Deskstar NAS drive and have now stuck it in the server. I had to disconnect the BD drive but it's not like that's actually been used for ages anyway.

Now when one of the 3TB drives kicks the bucket apparently the array will be rebuilt faster with the 6TB drive.

[Leon]

External drive works on my work computer, so I was able to retrieve everything from it, and put it onto my new WD drive.

[Yw-slayer]

One of the 3TB drives in the RAID array has died. Time to get another 6TB drive and stock that in.

[Yw-slayer]

What a surprise. A third one is now in its death throes.

This really may be the worst drive ever, save perhaps for the old Deathstars.

[Yw-slayer]

Windows Storage Spaces in Parity Mode is great. Pull out the defective drive while the computer is running (to make sure it's the defective drive), no problems. Shut down. Replace with new drive. Turn on computer. Add new drive to the pool and remove old drive from it. Array rebuilds. Worth the performance hit on writing.

[Yw-slayer]

Fourth one starting to die. After it goes then I'll only have one left. :lol:

Worst. Drive. Ever?

[thesameguy]

Seems like all things of a certain type having lived a certain life should fail at the same time, does it not? It's not a random failure, which sucks, but rather a known lifespan, which is good.

[Yw-slayer]

A substandard known lifespan.

[thesameguy]

How long have you been running them? Isn't that a three or five year old model?

[Yw-slayer]

Only around 3 years. I'll check HDD sentinel but they were bought fresh with the CPU which is a 3470, so notnmore than 3 years.

[thesameguy]

I think three years is all backblaze gets out of their drives. Seems about average IME.

[Yw-slayer]

I suppose. Let's see how the Deskstars go.

[thesameguy]

I've been buying those in 4tb lately as well. I think there are seven or so of them in production. They'll never get run to the end, they'll be replaced in two years, but I'm happy as long as they make it that long without a failure. ;)

[Yw-slayer]

Fair enough!

[Yw-slayer]

To try and cut a long story short:

1. I foolishly got lazy and didn't stick in the new Hitachi.

2. One of the two remaining Seagates is now at 9% health(!!) and Windows Storage Spaces has seen fit to "retire" both that and the other Seagate which is still at 100% health.

3. In order to substitute the new Hitatchi and remove either Seagate, I need 6 SATA ports. But my motherboard only had 5 left (since 1 is for the boot SSD) so I bought a Vantec 4-port SATA-3 PCI-E card (very cool and cheap on Amazon BTW. In fact, so cheap that I got 2 of them just in case).

4. However, now for some reason 1 of either the Seagate or the Hitachi drives will become invisible to Storage Spaces in Windows intermittently, causing the repair of the array to fail.

5. I suspect it's mainly to do with the PSU, perhaps too much load on 1 or 2 rails. But I thought that was sort of unlikely? I have no more than 2-3 drives on each cable. I can't find the manual and can't be effed to go figure out which cable goes where.

I suspect the easy way out is actually to build a new Skylake system - which I am going to do anyway - with 6x SATA ports (if I feel like it), stick an M2 SSD in, stick all the hard drives in, sort out the bloody storage spaces there, then stick them back in the server.

amirite?

[thesameguy]

Definitely a weird problem - modern HDDs really shouldn't be any sort of load on modern PSUs... somewhere in the neighborhood of 20-30w total split (albeit not evenly) between 5v and 12v. Not a lot of power. A 300w PSU should be able to run most any CPU at full tilt and still support six drives.

Is it possible you have a *flaky* PSU and that's what's causing the whole problem?

[Yw-slayer]

Well, I was about to tell you (as I knew you would) not to reply, but you beat me to it. I fixed it by:

1. Actually listening to the hard disks and noting that it was 1-2 of the Hitachis which were trying to spin up but failing.

2. Swapping the el cheapo Molex to 2xSATA connector attached to the PSU cable for the second Silverstone 1 SATA to 4 SATA connector that I'd also bought alongside the Vantec (the first one was also used for similar purposes in relation to the Seagates which had some clicking problems).

3. ???

4. PROFIT

[thesameguy]

Wha? Bad connectors? :lol:

[Yw-slayer]

Possibly. But overnight the repair stopped at 77%. I think something is overloaded and I can't find the other spare PSU cables. After the swap is done I'll remove the Seagates and hopefully it'll be fine. Not keen on looking for the PSU cables (although I'm sure I can find them) or, for obvious reasons, on buying another PSU.

[mk]

Is it the same PSU that powered those failed drives?
Its specs may be different now.

Any scraps around?
Put two cases side by side,
leave helper case motherboard connected and use its PSU for drives of other case.
That way the system should stay up long enough.

Also, use earthed AC plugs or bolt metal parts of the cases together.

[Yw-slayer]

It is, yes.

But I may have spoken too soon as when the repair gets to near the end then it hangs (although I can just about shut down Windows). Last time it got to 93%.

I suspect the HGST drives may be shutting themselves down as they end up with 59 degrees(!!) temp due to the copying. I think it's nearly done so I'll re-run it tonight, if it gets to 100% as I hope it does then I'll rip out the Seagates and the server should be fine for at least 2-3 more years.

I am also hoping that upgrading it to Win10 will resolve any minor issues, as that has in fact solved a very random "very slow to initially respond after logging in from desktop being at rest" problem on my main desktop (OCed 2600K).

[Yw-slayer]

SUCCESS!!! 100%.

Now to tear out all of the shitty Barracudas, make sure all of the HGST drives are installed behind a fan of some sort, and install Windows 10.

After I go to my wine cellar. Then wake up for gym.

[thesameguy]

Haha... just had my first 3tb fail at work last night. It had been on for three years, so whatever. I replaced it with another one because ZFG.

[Yw-slayer]

They will all die. By my hand, which creeps across the land.

[thesameguy]

I've got 13 more of them, 9 in production and four on the shelf.

DO YOUR WORST


(Seriously, because if we have to buy new drives I get credit card perks :lol: )

[Random]

$80-grand worth of cc perks... :lol:

[thesameguy]

Oh yeah... they took back $50k but left $30k. The guy on the phone who told me it was going to happen said he apologized but that was the final word.

[Random]

Oh yeah... they took back $50k but left $30k. The guy on the phone who told me it was going to happen said he apologized but that was the final word.

Well if that's how it has to be....

[thesameguy]

I feel very good about it. Because I could actually come up with the $30k as legit, the other $50k made no sense. I was actually relieved to get that call because it jived with what I had come up with. Anyway, it's sitting in a bank account doing nothing til the end of the year. Then I will buy all the remaining 3tb Seagate drives in the world and give them a happy, climate controlled home.

[Yw-slayer]

Don't you mean put them out on the lawn and then buy more 3tb drives to replace them when they fail?

[thesameguy]

I live in California, we don't get lawns anymore.

I leave shit in the dirt.

[Yw-slayer]

Yes. And then you will get more CC points.

[thesameguy]

HA!

HAHA!

The 3tb drive is fine. Zero SMART errors logged and passed a low level format just fine. It appears the issue was the RAID controller locking up and horking the drive erroneously. 3tb drive failure rate equals ZERO.

[Yw-slayer]

Well, in my case they were all hooked up to the MB SATA ports. I'd rather replace the drives than the MB anyway.

[Yw-slayer]

Time to get me mah money back, bo-yy!!

https://www.hbsslaw.com/cases/seagate

Or it would be, if they allowed submissions relating to overseas purchases.

[LHutton]

The 1TB SATA SSDs have come down in price, so they may be worth considering for a longer term, more reliable storage device.

[thesameguy]

Time to get me mah money back, bo-yy!!

https://www.hbsslaw.com/cases/seagate

Or it would be, if they allowed submissions relating to overseas purchases.

Oof. That's gonna be a rough case. I have mixed feelings about it as I don't like class action suits, but this does seem to have merit.

...Wonder how much I can get for the 14 I purchased? ;)

...Or how much I can get to keep quiet about my 0% failure rate. :D

[21Kid]

...so far. ;)

[thesameguy]

They will all be out of service within the next 3 months, so I think I'm safe. :)

[LHutton]

I reckon HDDs may be gone in 5 years time.

[Yw-slayer]

They will all be out of service within the next 3 months, so I think I'm safe. :)

Yup!

[balki]

750GB 2.5" HDD crapped out (from 2011?), haven't backed up in over a year.
How much will it be to ship to a lab and have them try to recover the data? $500+ is the only, rough quote i got

[Yw-slayer]

Doesn't sound too far off.

[thesameguy]

Depends on the nature of the failure. IME, typically around two grand.

[Kchrpm]

That's quite the range, and man, that's a lot either way. I don't feel so bad about not trying to recover my dead drives now.

[balki]

damn, i was hoping someone would come in and say 'send it to this place in China. $50 and a month later they'll do the job'
so, what's the easiest way to backup data offline (don't trust the cloud's security) on a laptop?

[Kchrpm]

You don't trust the cloud, but you do trust random IT place in China?

[balki]

I don't have $500-$2000 worth of data on the b0rked HDD, I don't have a choice.
something something ... the fappening?

[thesameguy]

The easiest, cheapest, most reliable way to back anything up in 2016 is the cloud.

Any other scenario you are buying additional storage, which you are then responsible for plugging in and updating periodically and keeping safe the rest of the time.

If you know precisely what you are backing up, the cheapest way to do it is an external drive and whatever software happens to come on it. Something comes on all of them.
If you don't know precisely what you are backing up, the most reliable backup will be an image-based back up like StorageCraft, Acronis, or Macrium and an externak drive.

$6/mo gets you 1tb of Onedrive space
$50 gets you a 1tb external drive
$50-$80 gets you image-based backup software

[thesameguy]

Side note, if you need what's on your dead drive, send it out for an examination. Worst case they come back with a number you can't handle. Best case they come back with a number you can. IME, for typical storage failures - which are electronic rather than mechanical or magnetic failures - $2k is the target. Maybe if you're lucky it will be cheaper.

[Yw-slayer]

:lol: Random IT place in China. No.

[LHutton]

They probably have a back up of your hard disk already.

[balki]

500, 2000, ....it's not worth it for some photos and one document.
I'm going to Walt Disney it and hope it'll be much cheaper in the future.
In the meantime, do I keep the HDD in the freezer?

[Yw-slayer]

I don't think a freezer would help much. I don't think hard drive repair will get cheaper in the future. USD500 is about what I was quoted in mid-90s to extract data from a 540MB (or maybe it was 1GB) hard disk.

It all depends on how important the photos and documents are to you. It doesn't sound like they're worth the money, so forget about it.

[LHutton]

Well the photos are probably his porn stash, so very important.;)

In the future with SSDs, will we ever have to worry about this again.

[thesameguy]

Fuck yeah, and SSDs make data recovery almost impossible. With magnetic storage the data exists regardless of the actual ability to get to it. With SSDs, the data might simply not exist. Backup and replacement cycle is WAY more important with SSD, and we don't have forty years of metrics to give us a clue what we're doing.

[balki]

recover ssd? ...
1587
the cloud it is

I don't think a freezer would help much. I don't think hard drive repair will get cheaper in the future. USD500 is about what I was quoted in mid-90s to extract data from a 540MB (or maybe it was 1GB) hard disk...
mid-90's? 500MB should have been the max around 95-96. Seems like back then sizes would double every 12 months though

[thesameguy]

Nonsense - the WD 1gb drive was available in '95.

[Yw-slayer]

Yes, I'm sure I had A 1GB Barracuda with scsi-2 (adapted 1540 for LIFE YO) in 1996 or so.

[LHutton]

1GB was serious stuff back then, many games were on a single floppy. Amiga 1200 had a 120MB hard disk and it seemed massive at the time with games on 880KB and 1.76MB floppies.

[LHutton]

Fuck yeah, and SSDs make data recovery almost impossible. With magnetic storage the data exists regardless of the actual ability to get to it. With SSDs, the data might simply not exist. Backup and replacement cycle is WAY more important with SSD, and we don't have forty years of metrics to give us a clue what we're doing.
But SSDs are several times less likely to give you that issue in the first place, because the part that normally fails isn't there to fail. But yes, always back things up.

[thesameguy]

I am not sure why you think that, but there isn't a single SSD technology that is mature enough at this point to develop any sort of real world measure of how reliable they are or how long they will last.

[LHutton]

There's 5-6 years worth of returns and refunds data. It's not 40 years, but it's good enough to provide a hint. Commonsense says, less moving parts, less failure.*

Plus nobody can usually be arsed to have a HDD restored anyway






*Excludes Windows 10.

[thesameguy]

Except that flash memory and controllers being produced even last year much less five years ago are not the same technology as the stuff being made now. Likening a c2010 Samsung MLC-flash based device to a c2016 TLC-flash based device is like extrapolating Ford Ecoboost reliability from a Thriftpower Six. That is, a waste of time. That is not true of mechanical drives, where much of the technology was static for years and it was just storage densities and costs which improved. Although not a direct comparison, it's close enough where things can be learned.

For example, when was the last time a mechanical drive spontaneously erased itself? When was the last time a mechanical drive caused a BSOD on a specific date? When is the last time a mechanical drive permanently lost speed because it couldn't delete files fast enough? These are all challenges SSDs have faced in the last few years that they no longer suffer from but which would dramatically affect warranty data. Huge recent issues that are meaningless now.

[LHutton]

Early SSDs had problems sure, but modern ones do not. And actually mechanical drives have different mechanical parts and some of them are shit, hence this thread.

[thesameguy]

No lie: The very last 3tb drive I had in production - in my own office computer - just bit the dust. I came in this morning and heard it making unusual noises, so I quickly copied all the data elsewhere and nothing was lost. Doing a diagnostic on it now, but it's definitely not happy! :lol:

[Kchrpm]

Time for a 5 TB SSD, amiright?

[Yw-slayer]

If only.

[thesameguy]

I replaced all the 3tbs - except this one - with 4tb HGSTs a couple years ago. And I just replaced those 4tb HGSTs with 6tb Seagates. So, this 3tb skipped the 4tb phase and went straight to six!

I left it doing a surface scan last night, and this morning it's completely dead. I got out in the nick of time! If it had started having problems tonight, I would have come back Monday to lost data. So pro.

[Kchrpm]

http://gizmodo.com/wow-samsungs-new-16-terabyte-ssd-is-the-worlds-largest-1723884725

Do it.

[thesameguy]

If it's really only $5k-$7k, I would probably buy one or two tomorrow. No joke.

[Yw-slayer]

In 10 years I bet there will be very few traditional HDDs in use by SRS BUSINESS users.

[thesameguy]

I get spammed daily by people offering multi-tb SSD mass storage solutions. SSDs have a lot of promise, but limited iops and uncertain long-term writability is a downer. It's fixable now, just none of the big guys are doing it so you get sketchy small businesses offering "solutions." I guess the big guys are enjoying the massive markups of SAS still. Very excited to see how m.2 affects all this - definitely not getting involved in a local company's SATA offering at this point.