[Gerrrt]

What changed

Docs only; nothing on smaug changes yet. ADR-0052 decides that at #558's disk swap, smaug's Exos pair moves from the MegaRAID 9340-8i to the chipset's free SATA ports and the card comes out. replace-the-nas-disk.md §5 and §6 are rewritten so the swap carries it out.

Why

Closes nothing yet. Refs #571, whose "done when" is met once the swap happens (§6 now closes #571 with #558). The pool has sat behind a MegaRAID in JBOD since the build, and its firmware's error path (60 s timeouts, aborts, an Online Controller Reset) is what ZVTBSDL3's fault ran through. Five of the chipset's six SATA ports are free.

• Chosen: chipset AHCI, card out. The only tray cable is the card's own mini-SAS HD breakout, so the move needs two plain SATA III cables, bought with the replacement drive.
• Rejected: IT-mode firmware, which would be a crossflash on the pool's only controller with no spare card.
• Fallback: if the drive isn't on ahci or the pool doesn't import after the move, the card and breakout go back in and stay as the arrangement by decision.

Blast radius

smaug and erebor only, and only when the runbook is run at the swap. §5 moves the surviving drive first, alone, on the one-legged pool. ZFS imports by label, moving a cable writes nothing, and the revert is to put the card back. §6 adds the SMART baseline re-check, because ADR-0047's /dev/sdc row can move once all three disks share one controller.

• No change to network segmentation or firewall rules
• No new port published to a VLAN that could not already reach the service
• No credential added outside secrets/*.sops.yaml

Verification

• scripts/lint.sh passes (markdownlint, yamllint, shellcheck, actionlint, editorconfig)
• scripts/check_docs.py: docs OK (README now counts 51 ADRs)
• Deployed to the lab and confirmed working (happens at the swap, per §5)
• Docs updated: ADR-0052, replace-the-nas-disk.md, hardware.md, README.md, changelog.md

🤖 Generated with Claude Code