SMART says my hard drive is dying, why did I have to check this manually?

[darxus2]

This hard drive has been flaky (randomly failing fsck, losing files).

"sudo smartctl -a /dev/sdc" says, among other things:


SMART overall-health self-assessment test result: PASSED

Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 199 193 051 Pre-fail Always - 329426
3 Spin_Up_Time 0x0027 156 141 021 Pre-fail Always - 9191
5 Reallocated_Sector_Ct 0x0033 157 157 140 Pre-fail Always - 344


The man page says:

'Each Attribute also has a Threshold value (whose range is 0 to 255) which is printed under the heading "THRESH". If the Normalized value is less than or equal to the Threshold value, then the Attribute is said to have failed. If the Attribute is a pre-failure Attribute, then disk failure is imminent.'

And it's way over all three of those pre-fail thresholds.

How should I have noticed this? Is the lack of notification a bug, or did I miss something?

[darxus2]

Looks like I mis-read the man page, specifically the part that says "If the Normalized value is less than or equal to the Threshold value".

[weatherman2]

If the RAW value of "Reallocated_Sector_Ct" truly is 344, then you've got a catastrophic problem or are about to have one. That means 344 sectors on the surface failed and were replaced with spares by the drive firmware. Once you run out of spares, the drive will be unable to replace new failed sectors and you will experience data loss, etc.

A few reallocated sectors is usually OK unless the number begins increasing regularly. 344 seems very bad to me.

What's the value of "Current Pending Sector Count" ? That's another one to watch. it should be zero; if not, then some sectors also can't be read already.

I would run the extended S.M.A.R.T. test immediately - use GSmartControl to do it with the Gui or figure out the options to smartctl to do it on the command line.