RunbooksOverview

Runbooks

Step-by-step fixes for incidents we’ve actually hit. Each one is written so someone who has never seen the problem can resolve it.

Writing a new runbook

When you resolve an incident, add a runbook while it’s fresh. Keep the shape consistent:

  1. Symptom — what the reporter sees.
  2. How to confirm — the exact command(s)/checks that prove the diagnosis.
  3. Root cause — what’s actually wrong.
  4. Fix — numbered steps.
  5. Prevention — what stops it recurring.