Operator Runbook

Operator Runbook

Back to Home

This is a short operating rhythm, not a substitute for host monitoring or the authoritative self-hosting docs.

After Every Deploy or Upgrade

1. Confirm /up and /operator load.
2. Verify the reported release and commit.
3. Check migrations, queues, the backup worker, scheduler, and Reverb.
4. Send a visitor message, reply as an agent, and confirm live or fallback delivery.
5. Review failed jobs and recent application/container logs.

Daily or Alert-Driven

• Investigate readiness warnings and failed queue jobs.
• Watch disk, database, Redis, and object-store health.
• Confirm mail and realtime provider errors are not accumulating.
• Review security-sensitive operator and break-glass activity.

Weekly

• Confirm scheduled backups completed and an offsite copy exists.
• Review storage growth, log retention, and attachment scan health.
• Check for dependency, image, host, and Wayfindr release updates.

Regular Recovery Drill

Provision a clean disposable VM, install the previous release when testing an upgrade, restore a real-shaped synthetic backup, upgrade, reboot, and repeat the support-loop smoke. Use scripts/smoke/public-artifact-reverify.sh after reboot when the VM was installed with a persistent evidence target directory. Record commands, versions, timestamps, and any repair made.

Use the full runtime contract and backup/restore guide when a check needs exact commands or recovery semantics.