[lug] Backup

[John Karns]

On Tue, 27 Dec 2005, Dean Brissinger wrote:

> On Tuesday, December 27, 2005 10:52 AM Zan Lynx wrote:
>
>> On Mon, 2005-12-26 at 08:23 -0700, Sean Reifschneider wrote:
>>> On Fri, Dec 23, 2005 at 02:39:01AM -0700, Nate Duehr wrote:
>>>> Always liked Seagate drives better anyway...  ;-)
>>>
>>> I recently tried some Seagate 300GB drives and out of 4 I had one
>>> fail in the first month...  Not overly impressed with those 300GB
>>> SATA drives.
>>
>> Infant mortality.  It's quite a common problem in electronics.  But
>> if it works past the first few weeks, it's likely good for years.
>
> Seagate has known problems with their big disks.  Something to due with
> a change in the bearing size and amount of lubricant that causes an
> internal ESD buildup.  This is a known problem that seagate hasn't been
> very willing to fess up to in the Fibre and SCSI market (all disks have
> the problem but only 10% expect failure?).  The bottom line is that
> under heavy write access the bearings break down and stall the platter.
> The drives can be power cycled a number of times as a temporary
> recovery.

Hmm.  Reminiscent of the "stitching" problem that a run of their product
had back in the days of the 40 MB drives, where one had to physically
shake/rotate the drive in the around the platter axis to unstick the
heads from the platter surface after cool-down.  Happened to me just
after the warrenty had expired, of course.


> I've also seen heat kill all brands of big disks.  A good chassis is
> worth the extra money.  Thus far my WD 300's are doing good with 24/7
> read/write activity after a month.  I stuck a big fan in front of them
> to blow the heat away though.

Any idea if their IDE series drives suffer from the same malady?  Several
months ago I picked up a 300 GB external USB model.  Overheating will
probably not be problem though, as the timeout for spindown after
inactivity is rather (too) short - about 5 min, and the thing spins down
until the next read / write request.  It would be nice to be able to set
it the timeout a bit higher - say about 15 or 20 min.

-- 
John Karns