[LUNI] Re: Testing IDE drives

[David Gibbs]

Keith T. Garner wrote:
> Surprisingly, SMART actually did work for me once.  My main server box at home
> started reporting "the drive will die within 24 hours" so I stayed up that
> night and transfered stuff over, and just as I finishing the transfer, it
> totally gave up the ghost.  So its not completely worthless.  :)

I don't rely on SMART ... I don't trust it.  More often than not, my
ears* have told me a drive is failing.  This is why I backup my critical
data to a tarball on a USB drive every night ... and then, on my
critical system, I dump the tarball to a DDS4 tape.

* For some reason, I've always been sensitive to failing mechanical
computer hardware.  About 25 years ago, when I was in my 2nd full time
job (working for a consulting company on IBM S/36), I went into the
clients computer room (a glorified filing closet) and heard a faint
whine coming from the area of the S/36's drive bays (fwiw: The drives
were 200mb with 6 24" diameter platters).  I told the clients MIS
manager, who went in but didn't hear anything.  He told me to call IBM
if I really thought there was a problem.  I called our IBM CE (we were
on a first name basis, for various reasons).  He was in the area, so he
would swing by just to check.  He came by, didn't hear anything either,
but checked the error logs, and found the drive was logging errors.
Suffice it to say, we were able to get a full backup and replace the
drive.  Our downtime was about 24 hours.  Had the drive failed, we would
have been down much longer (response would have been the same, but lost
data would have had to bev rekeyed).  Please keep in mind this was
before RAID was feasible.

So ends this trip down memory lane. :)

david