Skip to content

Incidents

An alert is what the engine detected. An incident is what you decide to tell people. They are separate on purpose: not every alert deserves a public incident, and an incident can exist for something no monitor saw.

The Incidents page
  1. Go to Alerting → Incidents and click New incident.
  2. Enter a title — it is what your status page shows.
  3. Pick a severity: Minor, Major, or Critical.
  4. Select the affected monitor.
  5. Click Create.

Creating an incident automatically posts a first Investigating update, so the timeline is never empty.

You can also start from an alert: open a row in the Alerts log, click Create an incident, and the title, severity and monitor arrive pre-filled with the alert linked to the incident.

Investigating Identified Monitoring Resolved History looking into it cause found fix deployed, watching back to normal archived, filterable 09:1209:2709:4110:03 Every update is timestamped and attributed, then pushed to the status page, e-mail subscribers and the RSS feed. Severity: minor · major · critical.

While resolving an incident, post updates — each one is timestamped and shown on the public status page:

  • Investigating — You’re looking into the issue.
  • Identified — The cause has been found.
  • Monitoring — A fix is being deployed or is being watched for regression.
  • Resolved — The service is back to normal.

Resolve is a shortcut for a final Resolved update. Every update goes out to your status page’s email subscribers.

The page has two tabs:

  • Incidents — what you declared, filterable by All / Active / Resolved. Each row expands to its timeline, where you add updates, resolve, or delete it.
  • Visitor reports — the problems your status page visitors flagged themselves (see Alerts from a status page), with a count badge for the unhandled ones. A report is a lead, not a confirmed outage: check the monitor, then declare an incident from it or dismiss it.

Deleting an incident removes it and its whole timeline, on the status page too — resolve it instead if it actually happened.

SeverityOn the status page
MinorDegraded performance
MajorDegraded performance
CriticalMajor outage — the page’s aggregate badge turns red

For planned work, use a maintenance window instead: it suppresses alerts as well as informing visitors. For a lighter message with no timeline, a status page announcement is often enough.