Components and incidents

Services

A service is something you promise to keep running — an API, a dashboard, a queue. Name services the way your users talk about them, not the way your architecture diagram does.

Incidents

An incident is a period during which one or more services were not operating as promised. It has a state, a start time, and a list of updates.

An incident stays open until you resolve it. Resolving it does not delete it — the record is permanent, and the history page keeps it.

Maintenance windows

A maintenance window is planned work. Announce it before it starts; the difference between planned and unplanned downtime is almost entirely whether you said so first.