Phase 7.4 — Storage burn-in and replacement

Validating the replacement drives before trusting them with anything.

Six enterprise SAS disks are tested as a set rather than one at a time — surface scan, soak, full parity build, and a health report per drive. Roughly a day of continuous work. Anything showing a defect is rejected before it ever holds data.

These are what replace the consumer drives whose thirty-second write stalls caused the outages that brought this whole phase forward.

No downtime — the testing runs inside the server emptied in 7.6.