Lynn McGuire wrote:

"What SMART Stats Tell Us About Hard Drives"
https://www.backblaze.com/blog/what-...rive-failures/

http://www.extremetech.com/computing...s-about-to-die

Cites BackBlaze data 2 years ago on 40,000 drives. The same SMART
attributes mentioned back then to monitor the health of an HDD are the
sames ones mentioned in their newer blog.

That SMART doesn't dictate reliable prediction of failure is probably
due to the leniency allowed to HDD manufacturers on how they report the
SMART values or the means they use to mask them.

When a non-zero Current Pending Sector Count value doesn't drop to zero
is when I get concerned that the drive can no longer mask its newly
detected bad sectors to reserved space on the HDD. There is no more
reserved space for the remapping so the bad blocks remain in use. The
reallocation of data to the reserved space can't happen until the data
gets read from the bad block to move it. If the bad block is so bad
that repeated read attempts fail to capture the data then the bad block
cannot be mapped (by moving the data to reserved space).