Automatic incidents

When a component is marked down, an incident opens by itself, and it resolves by itself when the component recovers. Webhooks and subscribers are notified at each step. Servers on a node that is down are covered by the node's incident, so one node failure does not open dozens.

You can turn this off with Open and close incidents automatically in Settings. The status of the components still changes; only the incidents are left to you.

Writing an incident by hand

Go to Admin > Status Page > Incidents and press New incident:

  • Title and Message. The message is plain text. It is shown on the page and sent to webhooks and subscribers.
  • Impact: minor, major or critical.
  • Status: investigating, identified, monitoring or resolved.
  • Affected components: the components the incident touches.

Then post updates as things move: pick a new Status and write a Message. Every update is added to the incident's timeline and sent out. Setting the status to resolved closes the incident. You can delete an incident from the list.

Open incidents appear at the top of the page. Resolved ones stay under Past Incidents for the last 14 days.

Maintenance

Use Admin > Status Page > Maintenance and press Schedule maintenance to announce planned work.

  • Title and Details.
  • Starts and Ends, in UTC.
  • Affected components.

During the window the affected components show as "Maintenance" and the time does not count against their uptime. If you pick a node, its servers are covered too. Upcoming and active maintenance appears in its own section of the page, with its window and affected components.

Maintenance sends a notification when it is scheduled and another when it starts.

Times

The times you type for maintenance are UTC. The page also shows times in UTC, labelled as such, so visitors in any country read the same thing.

Previous: Components · Next: Designs