Proving an RMA: what Node0 learned by doing it

Every return-merchandise process eventually meets the same trap. A manufacturer’s health check is written to describe the drive, not to predict whether it will survive the next hour of real use. A drive can pass it convincingly right up until it fails in production for the second time. Two did here: an SSD that passed SMART twice and failed in service twice, and a 12 TB disk whose own self-assessment read PASSED the entire way through 12,232 reallocated sectors and the storage-level inconsistency those eventually caused.

The fix that stuck was to stop asking the drive for its opinion and start measuring the condition the suspected fault actually needs. For SSDs, the tell turned out to be power-cycle count read against power-on hours. The condemned drive’s controller was resetting itself roughly once per powered minute, against well under once per ten hours on its healthy siblings. That difference is unmistakable while barely disturbing a health check. For hard drives it is a load-reproducing burn-in pass rather than an idle read. Both live in one tool with two tiers, a short pass for a spare going back on the shelf and a longer repeatable pass for a suspect, each capturing its evidence per serial number.

The second half of the lesson is organisational, and it surfaced on 20260916 when an inventory audit found five drives on the spares shelf that had already been claimed and were already dead, reading as available stock, because the records had never distinguished “physically present” from “usable”. A spares shelf that only grows and never gets audited will end up holding dead stock that looks exactly like good stock. The repair was a data model, not more careful handling. Every part now carries a dated disposition alongside its status, including the deliberate decision to run drives with no recorded warranty until they fail rather than file a speculative claim.