Palimsest discrepancy

John Morris jmorris at beau.org
Tue May 1 18:03:18 UTC 2012


On Mon, 2012-04-30 at 12:04 -0700, John Reiser wrote:

> How old is the drive?  That model number says 3-platter, 320MB, SATA.
> If it's three or more years old, then just replace it.  A new drive
> is $90 or less in US, and 500GB is available for the same price
> and same other specs (size, power, heat, performance, ...).
> The data is worth far more than that.

Dunno about that attitude being universal.  On my workstations I use NFS
homes so there is no data, only code.  I depend on SMART being right
enough so I can usually yank a failing drive before it goes so bad the
worker can't login and use the machine.  But I could care less about any
data on it, it is only a clone; because NFS root is just too much of a
performance hit compared to the cheapness of drives these days.  I try
to keep one spare workstation ready to drop in place of a failed unit so
downtime is minimal no matter what manages to go wrong with one.

Because a failure isn't a big deal and SMART is usually pretty good
about giving an advance warning I run em until they die.  In fact if the
failure is just a few bad blocks I often zero the drive (or run the
manufacturer's low level tool) and if the SMART warning goes away I put
it back in service and usually get a few more years out of em.  My
average drive age is easily over five years on workstations.  Still have
quite a few 40GB IDE drives in service.  A 500GB drive might be dirt
cheap now but a fairly complete Linux install still fits on a 40GB drive
so why throw out perfectly servicable equip?

If your data is worth anything it shouldn't be subject to loss of a
drive regardless of age.

I.    THOU SHALT MAKE BACKUPS
II.   THOU SHALT KEEP THY BACKUPS CURRENT
III.  THOU SHALT VERIFY THY BACKUPS

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://lists.fedoraproject.org/pipermail/test/attachments/20120501/8f6b1b23/attachment.sig>


More information about the test mailing list