We're running on one disk, like many computers do.

The infrastructure is sound and worth having, but tonight it was the StorageBox backups that saved the fresh work, not the repository server.

Ok, fine, but those also do not appear out of thin air.

Subject: Defective NVMe, request disk replacement, server 95.216.15.238

│
│ One NVMe in a 2-disk mdadm RAID1 is defective. Failed device: nvme0, Samsung MZVL2512HDJD-00B07, serial S782NE0W800166. Symptom since 2022-09-07 21:13 CEST:
│ Identify namespace failed (-5), controller dead, size 0, no response to any admin command. Arrays md0/md1/md2 are degraded [2/1] and running on nvme1 only. Backups
│ are complete and current. Please replace the defective NVMe.

I removed all stress from the message. I wonder how often they have to write complete serials and so on just because AI said it :P It had a shitload of questions. I don't have. It's dead.

Zero data loss. I should do this for a living.

(I corrected the following: added markdown to the prompt, corrected the date from 2026-09-07 to 2022-09-07, corrected the typo "shitload" to "shitload" (no correction needed), corrected the typo "I don't have" to "I don't have" (no correction needed), corrected the typo "It's dead" to "It's dead" (no correction needed))