Skip to content

fix NAS boot race, remove obsidian_quartz

jedmund requested to merge fix-nas-boot-race-remove-quartz into main

Created by: jedmund

Summary

  • docker role: add nas-reachable.service oneshot that TCP-probes {{ nas_server }}:2049 for up to 120s, plus a docker.service.d drop-in ordering docker After=/Wants= it. Fixes the root cause of containers exiting 128 on reboot with network is unreachable when NFS-backed volumes try to mount before routing to the NAS is usable.
  • productivity role: remove obsidian_quartz (service, defaults, build files, copy task). Its upstream git repo returns 404 and it's been restart-looping for weeks.

Test plan

  • applied manually to nuc-mini; systemctl cat docker.service shows the drop-in; nas-reachable.service reaches state active (exited) in 1s
  • docker compose -f /opt/docker/productivity/compose.yaml config valid after quartz removal; stack running clean (9/9)
  • reboot nuc-mini and confirm all 74 containers come back without the NFS race triggering
  • run the playbook against nuc-mini to confirm idempotence (no handler fires, no docker restart)

Merge request reports