Re: RAIDFrame and desktop drives; GreenPower

[George Michaelson <ggm%pobox.com@localhost>]

On Thu, 17 Jul 2008 08:59:21 +0300, Jukka Marin wrote:

> On Wed, Jul 16, 2008 at 01:18:57PM -0700, Carlos Linares wrote:
>> Also, has anyone tried out the new GreenPower Caviar drives?  How are
>> they?  Do they run cool?  Are they fast?  Are they quiet?
> 
> I bought 12 pcs of 1 TB disks (Digital WD10EACS Caviar GP) some 8 months
> ago. They are pretty quiet and run cool.  They are not the fastest disks
> in the world, but I think the low power consumption and noiseless
> operation is worth the small loss of speed.
> 
> However, the disks proved to be extremely unreliable.  One of the 12
> died within 8 hours (it started reporting its capacity as 0 MB).  Four
> disks developed bad blocks and SMART errors (some within 24 hours, some
> after a few weeks of operation).  The disks were installed in a
> Supermicro rack case with good ventilation and the disk temperature
> never exceeded 40 degrees C.
> 
> After a lot of thinking, I replaced all Caviars with Seagate Barracuda
> ES.2 ST31000340NS's which use more power, produce more heat - but are
> faster and have caused me zero problems.
> 
> I still think the Caviar is an interesting drive, but for some reason,
> they were the most unreliable drives I have ever used in over 20 years. 
> Older (and smaller) Caviars have been working well for several years, so
> I thought all Caviars would be alike.
> 
> Thanks to Areca RAID controllers, I never lost any data and was able to
> replace the Caviars with Barracudas one disk at a time on a live system.
> It took a few days to replace all disks and rebuild the RAID6.
> 
>   -jm


Unless they deliberately randomize the product coming off the production 
line, you'd be drawn to suspect some common fault in the diskset that 
your vendor bought. 

or... bad storage in the shelves before sale. or, bad treatment in 
transit.

did you forget to slow down over a speedbump coming back from the shoppe?

I really do mean this (ok. not the speedbump) because to have a cluster 
of failure like this without a known-bad international recall looks like 
a clustering anomoly which is best explained by them either being faulty 
make in a small batch, or else otherwise treated badly in common.

I had this with a bulk purchase of NCD Xterms back in the day, all the 
ethernets failed at the same time. Sometimes, it just happens..

-G