Skip to main content

Going to production checklist

Before swapping pat_test_… for pat_live_…:

Tokens​

  • Production PAT minted, stored in your secrets manager (not in git)
  • Per-token IP allow-list configured for non-local environments
  • Tokens scoped tight — least privilege, no admin:all unless really needed
  • Token rotation schedule set (90 days is a reasonable default)
  • Recovery codes for any 2FA-enabled account printed and stored offline

Idempotency & retries​

  • Every POST carries a fresh Idempotency-Key per logical operation
  • Client retries with exponential backoff on 429, 502-504
  • No retries on 400, 401, 403, 404, 422

Webhooks​

  • Endpoint registered with the right event filter
  • Signing secret stored securely (env var, not source)
  • Signature verification implemented; reject anything outside the 5-min timestamp window
  • Endpoint responds 2xx within 10 seconds (queue work async if needed)
  • Test the failure path: confirm we mark you status: failing after repeated 5xx and you get a notification

Rate-limit budget​

  • Your peak request rate measured in staging
  • You have headroom below the 60/min write and 300/min read defaults
  • If you need higher limits, open a ticket

Monitoring​

  • You subscribe to status.vps-server.host (or its RSS)
  • Your dashboards include request_id for every error response
  • Failed webhook deliveries surface in your alerting

Audit​

  • You know how to query GET /v1/audit-logs filtered by action and time
  • Sensitive endpoints (:revealPassword, /domains/{id}/epp) are reviewed regularly in your audit log