Incidents and maintenance

Incidents

An incident is a report about a problem, with a history. Automatic incidents open when a check fails twice in a row (see Components and checks). Everything else you report yourself under Dashboard → Services → your status page → Incidents with Report incident.

An incident has:

• Title and First message. The dashboard saves them as the German text, and English visitors see that text; you can only add a separate English version through the API.
• Impact: Notice, Degraded, Partial outage or Major outage.
• Affected components: which parts of your page the problem hits.
• State, which changes with every update.

The flow

An incident runs through four steps:

1. Investigating: you know something is wrong, not yet why.
2. Identified: you know the cause and are working on it.
3. Monitoring: the fix is in, you are watching whether it holds.
4. Resolved: everything works. The incident is closed.

Each step is an update with its own text and timestamp: on the incident you pick the New state, write the Text for your visitors and click Write update. Every update notifies your subscribers and webhooks.

You can correct title, impact and affected components later under Edit. That is not a new update and sends no notification.

Maintenance

A maintenance announces a planned interruption. Under Maintenance you click Announce maintenance and enter:

• a Title and optionally a Description,
• Start and End, in your local time,
• Affected components.

The status runs by itself: scheduled until the start, in progress until the end, then completed. If you need it differently, switch the status by hand, for example when the work is done early.

A running maintenance is shown at the top of your page. Your subscribers get an e-mail 24 hours before it starts.

While a maintenance is running, failing checks don't open an automatic incident. A server restarting for an update therefore doesn't cause a false alarm.