Pending and Reallocated Sectors
The two counters worth setting an alarm on.
01The two SMART counters worth setting an alarm on — and what they are actually counting
Most SMART data is noise. Temperature trends, spin-up time variance, load cycle counts — useful background colour, but rarely the numbers that precede a failure. Two attributes are different. If either of them is non-zero on a drive that held zero before, the drive is telling you something specific and urgent, and you should act on it today.
Reallocated sectors (SMART attribute 05) count the sectors a drive has already moved. When the head writes or reads a block and the result fails verification, the firmware quietly maps that logical address to a spare sector held in reserve — the reallocated sectors count goes up by one, and the bad spot is retired from service. A handful of reallocated sectors on an older drive is not automatically a crisis, but a count that is climbing is. Growth means the drive is consuming its reserve pool. Once that pool empties, the next bad sector has nowhere to go, and that sector becomes unreadable data.
Current pending sectors (attribute C5) are more urgent still. A pending sector is one the drive suspects is bad but has not yet been able to verify or remap. It is, in effect, unconfirmed damage: the drive tried to read the block, got a marginal or failed result, flagged it for re-evaluation, and is waiting for a successful write-and-verify to either clear the flag or trigger a reallocation. Until that happens, any file that maps to a pending sector may or may not be readable on any given attempt. The count can fluctuate — a successful read under a retry can clear it — but a non-zero pending count means there is data in genuine jeopardy right now.
The practical consequence of both numbers is the same: stop writing, make an image of the drive immediately, and work from that image. Do not run a file-system check on the original. Do not attempt repairs. The process of imaging a failing drive is designed precisely for this situation — a sector that is readable today may not be readable tomorrow, and a drive that has started reallocating can accelerate without warning.
One nuance worth keeping: imaging tools such as ddrescue will log which sectors could not be read, giving you a map of exactly which files to worry about once the copy is done. That map is far more useful than guessing.
What you should not do is watch the numbers and wait. A pending or reallocated sector is not a prediction. It is already happening.
| Attribute | What it is | What to do |
|---|---|---|
| 197 Current Pending Sector Count | Sectors that failed a read and are queued for reallocation. The data in them is not readable right now. | Image the drive before anything writes to it. |
| 05 Reallocated Sectors Count | Sectors already retired and replaced from the factory spare pool. | Record it, then image the drive before anything writes to it. A climbing count means the reserve pool is being consumed. |
| 198 Offline Uncorrectable | Sectors the drive could not read or write even during its own idle checks. | Treat as data loss in progress. Restore rather than repair. |
Current pending sectors (attribute C5) are more urgent still.