Skip to content

2026-07-05 — Sepia Phase2 Container Updates

Summary

Applied the first container update batch for Sepia and verified the updated services after fixing a Grafana data-permission issue.

Context

This work follows the active Sepia container update plan and depends on the earlier phase-1 healthcheck and hardening work to validate post-update health.

Changes

Change 1: Updated patch/minor release targets

  • Files:
  • /opt/compose.grafana.yaml
  • /opt/compose.borgmatic.yaml
  • /opt/compose.seafile.yaml
  • What: Updated Grafana, Borgmatic, and the MariaDB image used by Seafile to the planned release targets.

Change 2: Verified service health after updates

  • Files: live containers on Sepia
  • What: Pulled and restarted the updated services, then confirmed borgmatic, grafana, and seafile-mysql were running.

Change 3: Fixed Grafana data ownership

  • Files: /opt/grafana/grafana.db
  • What: Grafana restarted into a readonly-database failure because the database file ownership and permissions were too loose for the container’s migration step. I corrected ownership/permissions and restarted Grafana successfully.

Decisions

  • Applied only the planned low-risk updates in this batch rather than attempting the more disruptive major version upgrades yet.
  • Kept the Grafana fix operational rather than changing the compose file, because the issue was on the live data volume.

Issues Encountered

  • Grafana initially failed migrations with attempt to write a readonly database due to the live data directory/file permissions.
  • After correcting ownership and permissions on /opt/grafana, Grafana came up healthy.

Next Steps

  • Continue with the remaining container update targets in the plan, one at a time.
  • After updates, move on to the DNS migration plan in order.

Generated: 2026-07-05