Handbook

Operate 301 — Enterprise deployment checklist

Audience: security, platform, and release stakeholders signing off on a Fleet host. Outcome: explicit go / no-go items before exposing Fleet beyond a lab.

Updated

Go / no-go

# Item Owner Evidence
1 Bearer + bind policy matches threat model (Security) Security Config review + GET probe without token fails as expected
2 TLS terminates at Caddy / LB with valid certs (Caddy + systemd) Platform Browser + curl/openssl s_client
3 Docker isolation documented (socket perms, optional rootless/gVisor) Platform Runbook entry
4 Backups of fleet.sqlite + etc/ scheduled (Backup & DR) Ops Restore drill ticket
5 Remote update path understood (Upgrade) Release Dry-run on staging
6 Observability hooks in place (Observability) Ops Dashboard screenshot or alert definition

Operational readiness

  • On-call knows symptom → runbook mapping (Troubleshooting).
  • CHANGELOG host-operator notes for the target version are applied.

Rollback tested

  • You can reinstall N−1 bits or restore N data dir from backup per policy.

Post-deploy smoke

  1. GET /v1/version, GET /v1/health.
  2. POST /v1/jobs with kind: docker_argv trivial container — HTTP 201.
  3. GET /v1/jobs/{id} reaches terminal completed.