Field note: the "dying" SSD that wasn't

A server’s management console reported two alarming things at once: its SSDs showed only 1% to 3% endurance remaining, and its RAID-6 array was degraded. On the surface, that looked like failing drives and an urgent replacement order.

The endurance numbers were wrong

The drives weren’t the server maker’s certified models. They were third-party enterprise SSDs. The server’s RAID controller couldn’t correctly interpret the health (SMART) data these drives reported, so it displayed nonsense endurance readings. The drives themselves were fine.

Lesson: when a health number looks wildly wrong, check whether the controller actually supports that drive before ordering replacements. A drive the controller doesn’t certify can report misleading health data.

The degraded array was a loose drive

The degraded array had a simpler cause: one drive had become unseated in its bay. After reseating it, the controller saw it as a foreign configuration (a drive with RAID information it didn’t recognize as current). Importing the foreign configuration and starting a rebuild through the controller’s command-line tool brought the array back to health.

RAID-6 can survive two failed drives, which is exactly why a degraded array deserves prompt attention. It’s also why an unseated drive, rather than a dead one, is a relief.

What I took away from it

• Question readings that don’t make sense. Near-zero endurance on drives that were working perfectly was the clue.
• Physical checks first. Reseating a drive costs nothing.
• Understand “foreign configuration” before clicking through it, and have a current backup before any rebuild.
• Using non-certified drives has tradeoffs, including health reporting you can’t fully trust.