Is it a bad idea to buy 4 HDD's at once?

[Aetherone]

Originally Posted by Jadlak

Working for a number of years in an RA department for a retail store I am intimately aware of the 'bad batch' scenario.

Seconded. I've seen large raid-5 arrays where numbers of the drives fail sequentially. One drive fails, so it's replaced and the array rebuilt. A few hours/days later a second drive fails so it's replaced and the array rebuilt. A few hours/days later a third drive fails so it's replaced and the array rebuilt, and so on and so forth.

This gets a bit annoying when you have 12 or 16 or 24 drives in your array.

[hdkhang]

Originally Posted by Aetherone

Seconded. I've seen large raid-5 arrays where numbers of the drives fail sequentially. One drive fails, so it's replaced and the array rebuilt. A few hours/days later a second drive fails so it's replaced and the array rebuilt. A few hours/days later a third drive fails so it's replaced and the array rebuilt, and so on and so forth.

This gets a bit annoying when you have 12 or 16 or 24 drives in your array.

In this scenario it is likely that the cause of failure is not being addressed.

As for deathstars... my 10GB deathstar outlasted my 30GB seagate, 120GB seagate and 2 x 120GB WD drives. Being that it was the boot drive on one of my PCs for a long time meant it also got a lot of use. Right now it is sitting in my little cousin's PC that I built for them from old parts and it's still going, and they don't really care much for properly shutting down stuff... I attribute this entirely to luck.

[Aetherone]

Originally Posted by hdkhang

In this scenario it is likely that the cause of failure is not being addressed.

so what do you postulate as the cause?

[memnoch]

By the disks, but don't use them in production right away.

Set them up in a test box and run badblocks or something similar over the drives non stop for about 3-5 days straight.

This tends to sort out the ones that will die early, and the rest live on to have happy lives (usually anyway).

[hdkhang]

Originally Posted by Aetherone

so what do you postulate as the cause?

I don't work there, nor do I do hardware analysis... but if you are implying that the failure rate is due to an inherent "timebomb" in the drives from a similar batch instead of something external causing the issues... it's a bit farfetched.

[oh_noes]

Originally Posted by Aetherone

Seconded. I've seen large raid-5 arrays where numbers of the drives fail sequentially. One drive fails, so it's replaced and the array rebuilt. A few hours/days later a second drive fails so it's replaced and the array rebuilt. A few hours/days later a third drive fails so it's replaced and the array rebuilt, and so on and so forth.

This gets a bit annoying when you have 12 or 16 or 24 drives in your array.

Actually you're completely wrong. As per the rest of this thread, you're just a tin foil hat wearing hippie dumbass who should shut up and buy all 80 of his HDD's together.

Yeah thanks for the reply. I think I'll go two different brands (8 WD, 8 Samsung) with RAID6 and a cold spare. If it drops to DEFUCT (RAID5) I'll shutdown the array and replace the disk immediately, then rebuild the RAID6.

[oh_noes]

Originally Posted by memnoch

By the disks, but don't use them in production right away.

Set them up in a test box and run badblocks or something similar over the drives non stop for about 3-5 days straight.

This tends to sort out the ones that will die early, and the rest live on to have happy lives (usually anyway).

Already do it. Doesn't do much, all drives past the test. My drives have decent cooling so they rarely fail. I run linux format "drive test read/write" which takes many hours.

[one4spl]

Originally Posted by oh_noes

My drives have decent cooling so they rarely fail.

Aparently temprature (within reason) has little impact on drive reliability- http://research.google.com/archive/disk_failures.pdf

[Aratahu]

Ok, I just bought 3x WD10EVCS (1TB 24X7 rated drives) for a JBOD in my HTPC/home server. It already a WD10EACS; I'm taking out two smaller Samsung 0.5s.

I bought them more for the reason of having very low temperatures than outright performance; this fits the usage patterns well, in the Zalman HD160XT case they're going into I have a fan directly behind them but I do expect temps to be quite a higher than in e.g. my Antec 900 with low noise being a key target.

I debated getting a NAS device but seeing that it was about the price of those 3 drives alone I thought I'd try the JBOD approach. The HTPC is on 24x7 anyway, so might as well make the most of the juice it is sucking from the grid. (Seasonic S12+ PSU, so reasonably efficient).

[Aetherone]

Originally Posted by hdkhang

it's a bit farfetched.

Which is entirely why I'm more than happy to entertain alternative theories, of which you have provided precisely zero. Thanks for the contribution.

[oh_noes]

Originally Posted by one4spl

Aparently temprature (within reason) has little impact on drive reliability- http://research.google.com/archive/disk_failures.pdf

Google know a lot more than me, but I'm assuming that article was in respect to Data Centre cooling. At home, where most users don't even have decent case air flow, let alone air con, I think heat plays a bigger factor.

[hdkhang]

Originally Posted by Aetherone

Which is entirely why I'm more than happy to entertain alternative theories, of which you have provided precisely zero. Thanks for the contribution.

Lemme see, crappy controller, poor PSU, bad connector cables etc. Aging hardware can have all sorts of issues.

Its funny that you manage to rebuild the array before the next failure? Rebuilding arrays puts a lot of stress on the drives and as such you are likely to have a second drive fail during the rebuild process if the drive is about to die from being of a bad batch, that they do so a few days later suggests to me that something else is at play.