3 ms·
Although I don't think I'd ever DIY another NAS box, I would suggest the following, use a common chipset controller that exposes each drive/device separately wi
by tracker1 10y ago
Although I don't think I'd ever DIY another NAS box, I would suggest the following, use a common chipset controller that exposes each drive/device separately with all information, some don't. Also, prefer a system with ECC, Asus supports ECC on most AMD FX supporting motherboards for the less expensive option there. Also, have 2 spare drives on hand.
I went back to Synology after my uper-nas crashed and burned (really bad run of 3TB Seagate drives, 9 out of 12 dead in under 3 years)... Now debating between a 4x8TB or 5x6tb in a future nas box.
- ultramancool 10y agoThose 3TB Seagate drives are literally the worst in the business. The 2TB and 4TB are pretty good (roughly on par or better than competitors according to the backblaze reports) but those 3TB drives were utter shit for some reason, ~20% year over year failure rates. Some retailers have even started to sell the 3TB Seagate drives at almost the same price as the 2TB ones. Probably the only way to get rid of them.
- goda90 10y agoWell that's reassuring... I just built my home NAS. 3 3TB WD Reds and 3 3TB Seagate NAS drives. Two of the Seagates were DOA so I got them replaced. I ran full hard drive tests on them without problems but now I'm extra paranoid.
- ultramancool 10y agoThat's what RAID is for, just make sure you scrub it often enough to catch any potential failure. If you're ever looking for replacements, those 4 TB Seagates are hard to beat without paying enterprise-level prices, despite how much of a turn-off the high failure rate on their 3TB drives may be.
- CamperBob2 10y agoSafe to say the 3 TB drives are 4 TB drives that failed one manufacturing test or another.
- cyphar 10y agoPlease remember that back blaze was cracking open external hard drives in that case. Also, at the time, one of the large HDD manufacturers lost their main build plants. So who knows where Seagate got their components from in that 1-2 year period. Given the fact that Seagate's 4TB hard drives are considered by back blaze as the best HDDs, I find it a little odd that everyone assumes that manufacturing issues 4 years ago affect drives being built today. Also, back blaze statistics only apply if you're operating at back blaze's scale. If you're making a home NAS you would have to be _incredibly_ unlucky to hit a bad drive as quickly as back blaze did. I
- ultramancool 10y agoI've anecdotally experienced very high failure rates on them too, had 3/6 fail in about 5 years. And many other anecdotal reports indicate similarly bad results. Maybe not as bad as Backblaze but pretty damn bad. I absolutely agree though, the 4TB are a great improvement, you can't beat them for reliability unless you're really willing to pay for it. I've chosen them for my personal NAS and some servers even with great results so far.
- jzelinskie 10y agoI wanted to do something like this, but after reading about the failure rates of drives and raid, once one disk fails, it seems like you're fucked. Anybody got some stories or numbers that prove that while repairing my raid for one disk failure I won't just destroy the rest of my drives?
- zwerdlds 10y agoI've got a FreeNAS box which has endured two disk failures no problem. I have scheduled weekly rebuilds which I'm hoping finds bad blocks ahead of a failure-base rebuild.
- ultramancool 10y ago> I wanted to do something like this, but after reading about the failure rates of drives and raid, once one disk fails, it seems like you're fucked. Anybody got some stories or numbers that prove that while repairing my raid for one disk failure I won't just destroy the rest of my drives? Yeah, I've recovered several failed drives without issue, even on LVM systems where some partitions failed because they were striped but others worked because they were mirrored. No real issues overall. The only case where you might run into problems is when you get conflicting data between 2 drives, but I've never seen that happen in the real world. Some people will scare you into thinking that happens often but it hasn't been my experience. Often one drive just goes kaput completely or has an entire section which is completely unreadable or extremely slow, for which recovery is just a matter of marking that drive offline, swapping a new drive in and telling it to rebuild. The only case in which you lose everything is if you use RAID0. If you can do RAID10 that's the way to go IMO. RAID5 can be a little riskier because if one drive is damaged and one fails then you can wind up losing the bunch, but I use it on my personal NAS because it's mostly just TV and the important stuff is syncthing'd to my other machines. Regular scrubbing prevents these issues usually.
- pixl97 10y agoRAID6 is pretty safe. If you're using ZFS that's a RAIDZ2, you can lose 2 drives in a volume. RAID5 can be pretty risky with large drives that are the same type. >conflicting data between 2 drives, but I've never seen that happen in the real world Heh, run enough servers and you'll see everything eventually. Had a really fun one where a hardware raid card was telling us that writes were successful, but when data was coming back corrupt. It was writing bad data to two of the drives in the array.
- evv555 10y agoIf you have a way to deal with the noise then getting a 2nd gen rack server off ebay is a good way to go. At the least you'll get more ram and hard drive bays for your money.
- ta_donk_gt 10y agoI had a similar bad run with Seagate 3TB drives..3 out of 4 in 2 years. I was really thinking it was the NAS, but switched to WD Reds and have had no issues in 18 months.
- tracker1 10y agoThat isn't to say that reds/greens don't have issues... I've seen one of each fail, but they have been very solid compared to my experience with Seagate drives... the shame is that may very well be different today, but brand (dis)loyalty tends to carry for longer terms.