Quote:
Originally Posted by Carpetsmoker
SMART status is far from foolproof.
|
I'll agree, in so far as a
vendor normalized value is reported, and how that is converted into a "general health" summary by the electronics. But the raw values do not lie, and, of course, SMART is the only way to determine what the electronics on the disk drive knows about problems it has experienced.
I find the pinned sector counts and the detailed error logs extremely helpful. The offline tests, especially the long test, catch media trouble, including trouble with unallocated sectors. In RAID sets I can fail a drive with media trouble, clear the media problem (I use sysutils/e2fsprogs' "badblocks" tool for that), then reinsert the drive in the RAID set. The error logs, in particular, are good for root cause analysis; they can show you if an i/o failure is due to electronics, bus, media, or mechanicals.