Fjord Labs Engineering : Incident Runbook

Created by Alex Example, last modified on Sep 01, 2026

Declare an incident early. It is cheaper to cancel one than to start late.

First steps

  1. Open an incident channel.
  2. Page the on-call engineer (see On-call Rotation).
  3. Post status updates every 30 minutes.
kubectl get pods -n platform
kubectl logs deploy/api -n platform --since=15m

Escalation contacts

Escalate to the duty manager after 60 minutes.

Escalation matrix: escalation-matrix.csv

See also this page via a display link.

Attachments:

escalation-matrix.csv (text/csv)