Hitachi disk died at 28,000 h today

New to OpenZFS on OS X (Or ZFS in general)? Ask your questions here!

Hitachi disk died at 28,000 h today

Postby nekton34 » Mon Apr 11, 2016 9:09 pm

One of the four identical Hitachi 2 TB HDDs (manufactured Nov 2010) in my own disk array died at 28,000 hours (about 3.2 years) today. The raidz-1 remained functional and I was able to replace the failed disk and resilver without big problems (first time I've done this) following the WIKI here, so thanks.
The SMART utility shows all three other drives as PASS, but they are all at 28,000 hours too, so I guess it is time to replace them soon.
As a practical rule of thumb, is 3 years about the maximum to expect from consumer-grade HDDs?
Backblaze publishes disk failure stats for their server farms and HGST came out best I think I read recently. Is it better to spend more for server-grade disks or do they also start to fail around 3 years? Any thoughts or advice on brands/models?
nekton34
 
Posts: 34
Joined: Tue Feb 10, 2015 2:05 am
Location: Tokyo, Japan

Re: Hitachi disk died at 28,000 h today

Postby Brendon » Tue Apr 12, 2016 1:38 pm

I'd just keep a spare of whatever your favorite brand and size of HDD is, replace drives as they die. You never know how long these things will live for. If you replace all the disks with larger ones over time, when you replace the last, the raidz will grow to a larger capacity.

Cheers
Brendon
Brendon
 
Posts: 286
Joined: Thu Mar 06, 2014 12:51 pm

Re: Hitachi disk died at 28,000 h today

Postby incumbent » Mon May 23, 2016 11:14 am

That's great advice of course, but I have noticed a lot of people get really squeamish with raidz1 these days especially when block devices are larger than 1TB in size. I was traumatized enough that I only do mirrored zvols (and sometimes three device mirrors) or raidz2 pools now, but I wonder if on my workstation I'm not being overly cautious.

I wouldn't bother with enterprise drives @nekton34 — the warranty on them is better but in workstation quantity you'll have to do your own napkin math to see if it's worth it. Do I want nTB devices for five years? Probably not. When buying replacements and sizing up I generally just favor the manufacturer of the existing devices unless they're Seagate [1] but I do figure out the cost per gigabyte to make sure I'm not doing something completely stupid. Another thing about sticking with mirrors is that it's less expensive to expand storage simply because I'm buying fewer devices at a time but YMMV.

I keep at least one cold spare of the smallest size in my pool and sometimes experiment with warm spares in zpools. I've been using them in my FreeNAS successfully but don't bother on my workstation. Less disruptive to reboot my workstation rather than essentially bounce the entire house.

[1] [Seagate Paranoia]: Purely anecdotal on my part after having eight Seagate devices die less than 24 months of purchase I won't even bother RMA'ing them anymore because I spent too much time and effort dealing with them.
incumbent
 
Posts: 40
Joined: Mon Apr 25, 2016 8:52 am

Re: Hitachi disk died at 28,000 h today

Postby Brendon » Mon May 23, 2016 1:37 pm

I was able to expand the storage capacity of my array after replacing my last 2TB drive with a larger one. Turns out that at the moment Autoexpand does not work. Each and every disk device in the array had have 'zpool online -e' applied to it. Once done, it grew.

- BRendon
Brendon
 
Posts: 286
Joined: Thu Mar 06, 2014 12:51 pm


Return to Absolute Beginners Section

Who is online

Users browsing this forum: No registered users and 1 guest