How quickly does FreeNAS 'degrade'?

[sdspieg]

I just replaced a single defective HDD in my FreeNAS RAID, Everything was 'green' for quite some time. And when it started sending me warnings, I replaced the (supposedly) defective HDD. It resilvered nicely, but almost as soon as it did (within 2 days), 2 additional HDDs (apparently) became corrupt. Leading to a degraded pool. So here's my question. I like FreeNAS. As a backup system. And as a media system (sickbeard/couchpotato/plex/etc). But is it normal that hard drives 'die' after just a few (8-9) months? I now have about 5 'defunct' HDDs here, all of 3TB (some Seagate Barracuda, but also some 'NAS'-'grade'), and all presumably 'dead'. Is that normal? Do people really have to spend 6-700$ per years to keep their NAS going?

Thanks

 

[Bidule0hm]

A drive should last about 5 years (excluding infant mortality of course) so you definitely have a hardware problem like a bad PSU or insufficient airflow on the drives (who then run too hot) for example.

 

[jgreco]

In practice, drives die sooner than that in many cases.

The Seagate ST3000DM001's are in particular well known for failing. We saw a 50% failure rate (I say it that way 'cuz it sounds bad, but we only had 4 units). You can Google that along with "failure" for the Internet's depressing opinion of that drive. FWIW Seagate made a bunch of terrible 1.5TB drives too.

We've seen 3 of 13 with problems on the ST4000DM000's - one hard fail, two just developing bad sectors and not deemed bad enough to replace.

I agree with @Bidule0hm that you should check for any hardware issues, especially an undersized/cheap/etc power supply, but do be aware that it could just be the drives themselves. Yes sometimes they actually do fail like that. It's bad luck if so.

 

[AgentZero]

Some hard drives are better than others...and some die quickly - see these posts: https://www.backblaze.com/blog/best-hard-drive/ not to link outside of FreeNAS, but Backblaze has a poopTON of data on consumer drives and they are nice enough to aggregate it all together and give it away for free.

 

[jgreco]

No worries about the Backblaze link. Many of us here on the forum don't think a whole lot of the accuracy of their data, or some of their processes, but they easily and correctly point out that not all drives are created equal.

As with something like "Consumer Reports", the problem is that the information is only available on an "in hindsight" basis. It does no good to know that WD's last round of drives was very good if the ones going out the door right now are going to be a fail-o-rama. (Not picking on WD, substitute in Seagate if you want!)

 

[Z300M]

Some hard drives are better than others...and some die quickly - see these posts: https://www.backblaze.com/blog/best-hard-drive/ not to link outside of FreeNAS, but Backblaze has a poopTON of data on consumer drives and they are nice enough to aggregate it all together and give it away for free.

Interesting that Backblaze has data for a comparatively small number of WD drives -- and all are intended-for-NAS ("Red") models, whereas none of the Seagate drives are the intended-for-NAS models.

 

[jgreco]

not really that interesting. They usually seem to pick the cheapest drives available, as early as they are available.

 

[SirMaster]

Do people really have to spend 6-700$ per years to keep their NAS going?

Well if you buy disks with a 3 or 5 year warranty then you wont have to spend any money for that long other than maybe shipping when you RMA drives that have died.

 

[jgreco]

Well if you buy disks with a 3 or 5 year warranty then you wont have to spend any money for that long other than maybe shipping when you RMA drives that have died.

Riiiight, because you really want to put someone else's failed-then-"fixed"-and-recertified drive into your array... observation here is that recert drives are 2-3x more likely to fail.

 

[SirMaster]

Well, in my experience i've never really noticed any greater unreliability with the drives that I have gotten back from RMA personally. But if they were to fail again, I'll just RMA again.

My sample size probably isn't big enough, but I currently have 42 drives in operation at home and a decent number of which I have RMA'd over the years.

 

[jgreco]

Well, in my experience i've never really noticed any greater unreliability with the drives that I have gotten back from RMA personally. But if they were to fail again, I'll just RMA again.

My sample size probably isn't big enough, but I currently have 42 drives in operation at home and a decent number of which I have RMA'd over the years.

We sure have noticed. Refurbs are great for offline storage or gifts.

 

[SirMaster]

OP is talking about a home media storage system as am I.

You seem to be talking about a business storage system with more important uses.

It's a different situation that requires different decisions IMO. I feel like this disconnect happens a LOT on this forum.

 

[Z300M]

Riiiight, because you really want to put someone else's failed-then-"fixed"-and-recertified drive into your array... observation here is that recert drives are 2-3x more likely to fail.

I realize that hard drives are electro-mechanical devices and not purely electronic, which may explain the difference between your experience with hard drives and the suggestion I read years ago that electronic devices tend to either fail very early or last for years and that a refurbished/recertified device has in effect undergone more extended pre-shipment testing than a straight-off-the-production-line one and is past its early-failure phase.

I've had two Seagate drives replaced under warranty, and both are still fine -- one after 2+ years, the other after just a few months. Both replacements were of current models, the original failed models no longer being in production.

 

[jgreco]

I'd bet we've had over a hundred drives replaced under warranty over the years, who the hell counts. There's a box in the closet of failed drives that happened out of warranty too. The replacements are never as good as the new ones. Infant mortality among new drives can be counteracted by returning for a replacement disk, which is also brand new.

 

[gpsguy]

While I don't have empirical evidence to support it, it seems like consumer SATA drives produced after the Thailand floods have had a shorter life expectancy than those produced before the floods.

Sure, there have been bad combinations of vendor/model numbers, like the 1.5TB Seagates jgreco mentioned.

I've still got a bunch of *old* IDE/SATA drives of various sizes that are still functional.

 

[jgreco]

Sure, and I've had many drives that have lasted ten or even twenty years too. That tends towards being the exception, though.

 

[cyberjock]

OP is talking about a home media storage system as am I.

You seem to be talking about a business storage system with more important uses.

It's a different situation that requires different decisions IMO. I feel like this disconnect happens a LOT on this forum.

Yes, but no. Many users that are doing home media are very fickle with their storage because they have zero backups. So yes, you can argue that the disconnect happens alot, but much of the time it's worse than being in a business because there are no failsafes if things go horribly wrong.