How to Read a SMART Report Before Buying a Used Hard Drive

Every hard drive keeps a running diary of its own health called SMART (Self-Monitoring, Analysis and Reporting Technology). When you buy a used drive, that diary is the difference between a bargain and a headache — if the seller shows it to you, and if you know how to read it. Every drive we sell at Atomic Racks ships with a printed certificate summarizing exactly these numbers, so this post doubles as a guide to reading our own paperwork.

The five attributes that matter most

ID 5 — Reallocated Sector Count

When the drive finds a sector it can no longer write reliably, it retires it and remaps the data to a spare. A small handful of reallocations on an old drive isn't automatically fatal, but the number should be zero on anything we'd certify — and more importantly, it should be stable. A drive that reallocated 10 sectors three years ago is a different animal from one that reallocated 10 sectors last week.

ID 197 — Current Pending Sector Count

Sectors the drive suspects are bad but hasn't yet confirmed. This is the scariest attribute on the list because it describes the present, not the past. Any non-zero pending count means data the drive isn't sure it can read back. Walk away.

ID 198 — Offline Uncorrectable

Sectors that failed during the drive's own background checks. Like pending sectors, anything non-zero is disqualifying for a drive you plan to trust.

ID 9 — Power-On Hours

How long the drive has actually run. Enterprise drives are engineered for 24/7 duty, so hours alone don't condemn a drive — a datacenter pull with 30,000 smooth hours and clean health attributes is statistically a known-good unit that survived infancy. What you want is consistency: hours that match the seller's story.

ID 199 — UDMA CRC Error Count

Errors on the cable between drive and controller. A non-zero count usually indicts a past cable or backplane, not the drive itself — but it should not be climbing in the drive's current home.

Raw value vs. normalized value

SMART reports show a confusing pair of numbers per attribute: a normalized value (usually starting at 100 or 200 and counting down toward a failure threshold) and a raw value (the actual count). For the attributes above, read the raw value. For temperature (ID 194) and the vendor-specific attributes, the normalized value is often more meaningful.

What about SAS drives?

SAS drives don't expose the familiar 1–255 attribute table at all — they report through log pages instead. The equivalents to watch are the grown defect list (analogous to reallocated sectors), non-medium error count, and the read/write error counter logs. A clean SAS drive shows a short (ideally zero-length) grown defect list and no uncorrected errors.

The one test a listing photo can't fake

A screenshot of a SMART table only proves the drive was alive once. A completed SMART extended self-test (also called a long test) proves the entire surface was readable end-to-end at a specific point in time — it takes hours on a large drive, which is exactly why most sellers skip it. We run one on every drive before it's listed, and the result (with completion timestamp) is on the certificate in the box.

Quick checklist

• Reallocated (5): 0 preferred, stable if not
• Pending (197): must be 0
• Offline uncorrectable (198): must be 0
• CRC errors (199): explainable and not climbing
• Extended self-test: completed without error, recently

Browse our tested enterprise drives — every one ships with its SMART certificate and a 90-day warranty. Questions about a specific report? Send it over; we read these all day.