Runbook:

<One sentence: what is broken from a user’s point of view when this page is open, and which alert or symptom brought you here.>

Symptoms

• <What the pager, a user, or a dashboard shows: the alert name, the error code users get (for example 503 no_capacity), the panel that turns red.>
• <What kubectl get pods -n <system> or /readyz shows.>

Diagnosis

1. <The first command to run, exactly, and what a healthy and a broken output look like.>
2. <The next check, chosen by the output of the previous one.>

Mitigation

1. <The fastest safe action that restores service, exactly (a rollback, a drain, a scale-up), and how to confirm it worked.>
2. <What not to do, and why (for example: do not delete the WAL PVC).>

Follow-up

• <The fix that removes the cause, and where it is tracked.>
• <Links: the alert rule, the dashboard, the ADR or chapter that explains the component.>