Live platform monitoring

Every service on this page is checked every 60 seconds from an independent monitoring system hosted outside the infrastructure it watches. That separation is deliberate: a monitor living on the server it checks goes down with it, and stops reporting exactly when it matters most.

The check is not a simple ping. Depending on the service we verify that the port answers, that the certificate is valid and that the application returns the expected content, not just that the machine is powered on. A server that answers with a 500 error is listed as down, because for whoever uses it, it is.

The percentage next to each service is uptime over the last 24 hours, computed over the checks actually performed. The bars show the most recent checks in chronological order: each one is an attempt, and its colour shows how it answered.

What happens during an incident

The alert reaches the on-call team immediately, through several channels at once. If the affected service has automatic remediation, the system first tries to restart it and check again; if that is not enough, a person steps in. Incidents affecting customers are published on this page, above the list.

Maintenance windows

Planned work that may involve downtime is announced in advance and carried out during low-activity hours. During a window the service shows as under maintenance, which is a different state from down: it is planned and has an end time.