Records the 2026-07-20 incident by verbatim symptom: config vs running state
disagree, MAAS's own service_set reports healthy, and the region journal
shows Temporal 'Not enough hosts to serve the request'. Check is pgrep -c
dhcpd on EVERY controller; fix is snap restart maas on the region, verified
behaviorally. Includes the scope trap that made it invisible for days: an
already-built site keeps working because deployed VMs hold their addresses,
so it only surfaces when new nodes try to PXE.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KiUu1oqt76tWvV4vEC3NAr