Gentoo Archives: gentoo-user

From: John Runyon <me@×××.im>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] Hard drive noise
Date: Sat, 19 Dec 2015 03:48:37
Message-Id: o1tb3t8ydc2h5q8yc6nlustl.1450496866478@email.android.com
1 > The thing is, EVERY
2 manufacturer has had drives like these.  Well, the Hitachi drives
3 Backblaze goes on about would be an exception
4
5 you've clearly never heard of DeathStars if you think Hitachi is an exception...
6
7 -John
8 Sent from my phone
9
10 Rich Freeman <rich0@g.o> wrote:
11
12 >On Fri, Dec 18, 2015 at 7:49 PM, Dale <rdalek1967@×××××.com> wrote:
13 >>
14 >> I also found this after the reply from Ian.
15 >>
16 >> https://www.backblaze.com/blog/3tb-hard-drive-failure/
17 >>
18 >> No wonder they had it on sale. Heck, why didn't they just say it was a
19 >> good door stop instead of a hard drive???
20 >>
21 >
22 >Yeah, the only reason I'd want to use that model drive is if I had a
23 >raid5 composed of entirely different drives and for some reason the
24 >discount on the Seagate 3TB drive let me bump it up to a raid6 (and to
25 >be sure I'd never put more than one of those in an array). It is
26 >basically a doorstop.
27 >
28 >I had two of those go in the span of a year. One was replaced under
29 >warranty. The next was the warranty replacement. That one was no
30 >longer under warranty, but after a scathing Amazon review Seagate
31 >actually commented on the review asking me to contact them about a
32 >replacement. I didn't bother - I really was tired of swapping out
33 >drives at that pace and didn't consider the considerably-higher risk
34 >of a double failure worth it.
35 >
36 >i'd have to check - I think I picked a 4TB Seagate NAS drive to replace it.
37 >
38 >Somebody suggested not buying Seagate. The thing is, EVERY
39 >manufacturer has had drives like these. Well, the Hitachi drives
40 >Backblaze goes on about would be an exception, but they're
41 >SIGNIFICANTLY more expensive and I don't think it is worth the premium
42 >in a RAID. For a single-drive system I'd strongly consider them. I
43 >think I heard they were bought out at some point, so we'll see if
44 >their reputation holds.
45 >
46 >And that's the thing with brand reputations. These days MBAs milk
47 >reputations. Some finance guy realizes that people will buy this
48 >year's drives based on last year's reputation and cuts some corners
49 >and collects a huge bonus. Three years later everybody is dealing
50 >with drive failures. Every vendor does it. That one Seagate model
51 >was about the worst I've personally seen, but who knows what model is
52 >being sold today that in three years will turn out to be just as bad,
53 >and it could come from any of the vendors.
54 >
55 >I do try to look at the Backblaze stats for what they're worth, but I
56 >think the general advice applies well. Make sure you have an
57 >appropriate level of redundancy and backup strategy. Make sure to mix
58 >models of drives in your RAIDs. The whole point of a RAID is to keep
59 >the price down by increasing your tolerance of failures.
60 >
61 >And the whole NAS drive firmware thing really bugs me because they
62 >charge a premium for a few bits in flash memory that should be
63 >user-configurable anyway. Some of those drives have better vibration
64 >resistance, which bugs me less. However, the bottom line is that they
65 >probably will improve your RAID performance in the event of a failure,
66 >and they probably do tend to cut the corners less on them. But who
67 >knows, maybe the drive that fails next year will be the super-premium
68 >edition.
69 >
70 >All of this goes to one of my drivers for using btrfs (and in this
71 >regard zfs will do just as well). The checksumming means that I'm not
72 >really trusting the drive or its firmware at all, and I scrub my
73 >arrays weekly.
74 >
75 >Sorry you ended up with a bad drive... That model IS considerably
76 >cheaper than most of the others...
77 >
78 >--
79 >Rich
80 >