Compaq Smart Array 3200 Controller as a SCSI Controller - test-drb@ccmp.vtda.org

18 Jul 2020

On 7/17/20 7:07 PM, Paul Koning wrote:
...
  Yes, if you define it that way then clearly I agree. 
The thing is 
 that in most people's definition, "drive failure" means "the drive 
 is a door stop". 
Ya....  I've had too many "but the drive isn't a brick ... how could it 
be the failure" experiences to use that as my benchmark.  Now, if the 
drive is not doing what it's supposed to do in any (reproducible) 
manner, I consider it a failure.  Well ... almost any reproducible manner.

...
  And in fact, hard read errors are normal.  Every drive
has a spec for 
 the probability of that happening, and given the per-sector failure 
 probability and the sector count, the probability of SOME sector 
 failing to read when you read the whole drive is nowadays somewhere 
 around 1. 
Ya.  That's where the reproducibility of any given failure comes into play.

-- 
Grant. . . .
unix || die