I currently have a server with 32 bays, all are populated with 3TB drives - most are Seagate ST3000DM001s.  Despite what BackBlaze seems to have experienced, the drives do not burst into flames when you look at them.  Many of them have been in service for 4-5 years now, and my failure rate has been nowhere near what BackBlaze has published.

 

I do have a frequent "Timeout" problem with them, though.  Occasionally, they'll be dropped from my 3Ware RAID arrays, requiring a full rebuild.  After a drive has done this a couple of times, I tend to replace it (with a proper NAS drive).  As a result, I've amassed a collection of drives that I don't trust in a hardware array, but they have good SMART numbers, and pass the "long tests".

 

I was wondering: are these drives likely to cause problems in an UnRAID array?

 

Thanks in Advance!

Sounds like you have experienced vibration issues, which is typically added to "NAS" drives, but not in desktop drives. Too much vibration and the drive can not complete seek/read operations in a timely manner. Interesting video is available of slowing disk drives by shouting at them

 

Yes, these drives are not likely to cause the same problems with unRAID.

 

It is important to note that Backblaze does NOT use array controllers like the 3Ware, thus does not have drop out and rebuild problem.

Yes, these drives are not likely to cause the same problems with unRAID.

 

Thanks!

 

Your answer was what I was hoping for.

 

Sadly, I just discovered that UnRAID does not support multiple parity.  If I can get past that, I can see myself taking the plunge.  I don't know that I can trust single parity for an array with more than a couple drives, though.  (My current setup uses 8-drive arrays, and there's been a couple of times when RAID-6 has saved me in the past.)

 

EDIT:  I just saw that Dual Parity is being added in 6.2. 

 

It is important to note that Backblaze does NOT use array controllers like the 3Ware, thus does not have drop out and rebuild problem.

 

I have absolutely no idea what BackBlaze does to those poor drives.  Someone else argued (and I'm inclined to agree) that Seagate would be out of business if their hardware really was that unreliable.

 

Yes, these drives are not likely to cause the same problems with unRAID.

 

Thanks!

 

Your answer was what I was hoping for.

 

Sadly, I just discovered that UnRAID does not support multiple parity.  If I can get past that, I can see myself taking the plunge.  I don't know that I can trust single parity for an array with more than a couple drives, though.  (My current setup uses 8-drive arrays, and there's been a couple of times when RAID-6 has saved me in the past.)

 

EDIT:  I just saw that Dual Parity is being added in 6.2. 

 

It is important to note that Backblaze does NOT use array controllers like the 3Ware, thus does not have drop out and rebuild problem.

 

I have absolutely no idea what BackBlaze does to those poor drives.  Someone else argued (and I'm inclined to agree) that Seagate would be out of business if their hardware really was that unreliable.

 

I'd also say that because of the way unRAID works makes it less likely for two failures at the same time to occur. The only time all disks need to be spun up at the same time is during parity checks and rebuilding a failed disk. Otherwise only the disk being read or written to needs to be spun up. Allowing the drives to stay spun down the vast majority of the time increases their longevity.

Yes, these drives are not likely to cause the same problems with unRAID.

 

Thanks!

 

Your answer was what I was hoping for.

 

Sadly, I just discovered that UnRAID does not support multiple parity.  If I can get past that, I can see myself taking the plunge.  I don't know that I can trust single parity for an array with more than a couple drives, though.  (My current setup uses 8-drive arrays, and there's been a couple of times when RAID-6 has saved me in the past.)

 

EDIT:  I just saw that Dual Parity is being added in 6.2. 

 

It is important to note that Backblaze does NOT use array controllers like the 3Ware, thus does not have drop out and rebuild problem.

 

I have absolutely no idea what BackBlaze does to those poor drives.  Someone else argued (and I'm inclined to agree) that Seagate would be out of business if their hardware really was that unreliable.

 

I'd also say that because of the way unRAID works makes it less likely for two failures at the same time to occur. The only time all disks need to be spun up at the same time is during parity checks and rebuilding a failed disk. Otherwise only the disk being read or written to needs to be spun up. Allowing the drives to stay spun down the vast majority of the time increases their longevity.

on pure (hardware at least) RAID's, TLER is the important part of hard drives.  ST3000DM's (or any consumer drive) are NOT designed to be on a hardware RAID, and will suffer an unusually high rate of failure.  Hence why WD and Seagate both came out with their NAS versions which are basically a consumer drive with an enterprise firmware (supporting the TLER).

 

The actual instances of 2 drive failures on unRaid happening simultaneously (or during a rebuild) are exceedingly rare, and when they have happened, 99% of the time its probably been caused by accidentally knocking loose sata and/or power connectors to the drives - mitigated by using hotswap bays - a must have

My thanks for all of the replies... I'm feeling a lot more confident now (though I'll almost certainly continue using dual-parity).  I currently have 2 parity drives for each 6 data drives.  I'm always amazed that people can go way beyond my 1:3 ratio and still sleep at night.

 

I'm painfully aware of the TLER issue with consumer drives, and I'm convinced that those have been at the heart of most of my problems.  It's especially heartbreaking because it's a firmware issue, and should cost absolutely nothing to "fix".

 

I guess that if I have one concern now, it's the statement that I'm likely to have less problems because the drives spin down when not in use.  In fact, I've wondered if the high failure rates at BackBlaze has been due to the regular spin-ups that my drives rarely perform.  I'm not ready to live with a 25% drive failure rate!

I guess that if I have one concern now, it's the statement that I'm likely to have less problems because the drives spin down when not in use.  In fact, I've wondered if the high failure rates at BackBlaze has been due to the regular spin-ups that my drives rarely perform.  I'm not ready to live with a 25% drive failure rate!

If anything, just in theory drives (or anything in the world for that matter) will have less problems if you don't spin them down.  Spin ups / power on's are the hardest on anything's electronics / mechanics.

 

Backblaze reports should be taken with a grain of salt.  Yeah, they are a nifty thing to look at, but what they are basically reporting is akin to publishing the reliability reports of a Pinto competing in the Baja races.  Sure it'll give a guideline of just how well made a Pinto is, but its not something anyone in their right mind would actually do.

Also keep in mind that unlike most other RAID arrangements, you don't lose your entire array with a multi-drive failure in unraid. A 10 disk array using single parity isn't going to lose all 10 disks once you've exceeded unraid's single-failure tolerance.

 

That still might be too much of a risk for you, but it does change the equation a bit.

Backblaze reports should be taken with a grain of salt.  Yeah, they are a nifty thing to look at, but what they are basically reporting is akin to publishing the reliability reports of a Pinto competing in the Baja races.

 

I agree - but only partially.  I can understand timeouts, slower performance, etc, but failures are another matter.  I don't feel like I should spend my time worrying about an invisible line that I dare not cross... the line between home and professional is exactly that.  I don't really feel that it's unreasonable to expect the vast majority of consumer-level hard drives to take a three year beating without failing, especially when many of the competitors seem to be managing to do so.

 

Also keep in mind that unlike most other RAID arrangements, you don't lose your entire array with a multi-drive failure in unraid. A 10 disk array using single parity isn't going to lose all 10 disks once you've exceeded unraid's single-failure tolerance.

 

That does help... a bit... I think.

 

It should Especially since unRaid is not a backup strategy. So you should be backing up your really important data:

 

- on another disk in the array

- on another array in house

- on another PC in the house

- on an external drive (ideally stored in another location)

- on another array in another location

- on a crashplan-like setup or cloud drive

- etc ... the options are limited only by your imagination, and your tolerance for risk (where Risk = magnitude of loss * likelihood of event) ... use multiple for even more piece of mind.

 

and if you aren't keeping backups remember that ransomware WILL hit your shares so you are not just at risk for drive failures!!! A thread on the topic with ideas https://lime-technology.com/forum/index.php?topic=47961.0

 

and if you aren't keeping backups remember that ransomware WILL hit your shares so you are not just at risk for drive failures!!! A thread on the topic with ideas https://lime-technology.com/forum/index.php?topic=47961.0

 

I actually do do backups, though my data is in flux pretty regularly (mostly additions - rarely changes), and I haven't nailed down a strategy where everything is backed up all of the time.  I actually started a separate thread about my tape drive a few minutes ago.

 

Ransomware might be just about the only worry that I don't have.  All of my media is stored on one server and none of the normal user accounts can log onto it, or gain write-access to the shares.  I can't say it's impossible that it'll ever be breached, but I think it's less likely than a fire.

 

I guess a "perfect" strategy for me would be a way to do incremental backups since my last tape backup (without actually using tapes).  I'm sure there's a great approach would allow me to do nightly cloud backups of everything that isn't archived, but I haven't worked it out yet.

 

 

ahh great to hear. My job is done ;-)