[smartmontools-support]attributes and health check

[Ralph Blach]

As I understand the article in the Linux Journal, one does not use the
Attributes to determine the health of the drive, but on uses selftest
and health status, to determing the health of the drive.
When the health of the drive comes back bad, its time to replace the drive.

Is this the correct assumption?


Thanks

Chip

[Bruce Allen]

Hi Ralf,

Post by Ralph Blach
As I understand the article in the Linux Journal, one does not use the
Attributes to determine the health of the drive, but on uses selftest
and health status, to determing the health of the drive. When the
health of the drive comes back bad, its time to replace the drive.
Is this the correct assumption?

Yes, roughly this is right. However the health of the drive is related to
the Attributes, so one can often infer something useful about the health
of the drive by looking at the Attribute values.

Example: Maxtor disks typicallly have 500 spare sectors which they use to
replace failed disk sectors. You can monitor the number of sectors that
have been reallocated from the Attribute values. When this number gets to
500 the disk health status will fail because there are no remaining spare
sectors -- so it's nice to be able to see when the number of reallocated
sectors gets to 450.

Attribute values often tell you other useful things. For example when the
UDMA CRC error count is nonzero, or increasing, this indicated problems
with the disk cabling (poor connections, electrical noise, damaged cable,
etc). And so on.

To learn more, I suggest you study the example smartctl output on the
smartmontools web page.

Cheers,
Bruce

[PS: IBM disks have some of the BEST documented SMART data. IBM's OEM
disk manuals list the meanings of the different Attributes. So you can
also talk to some of your colleauges at IBM/Hitachi.]