On-call rotation
The rotation#
Fill this in: how long a shift lasts, when it changes hands, and where the live schedule is. Name roles rather than individuals, so the page stays correct as people join and leave.
| Role | Responsibility |
|---|---|
| Primary | Acknowledges alerts, runs triage, communicates status |
| Secondary | Backs up the primary when they do not respond or need hands |
| Escalation | Decisions with a business cost: rollback, data restore, public statement |
What you owe while on call#
- Be reachable, and able to reach a machine within the response time your team has agreed.
- Acknowledge alerts even when you already know the cause. Acknowledgement is how everyone else learns someone is on it.
- Keep a running note of what you tried. It becomes the incident timeline for free.
What you do not owe#
You are not expected to work a full day and then a full night. If an incident takes your evening, say so and take the time back.
Teams that skip that last paragraph pay for the outage twice: once when it happens, and again a month later when the person who absorbed it quietly stops volunteering.
Handover#
At the end of a shift, hand over in writing: anything still open and what you would try next, alerts that fired and turned out to be noise, and any temporary change still in place — a disabled job, a raised limit, a flag flipped.
An empty handover is a valid handover. A silent one is not.