How SpaceOS is running, right now

Probed independently, around the clock

A self-hosted Uptime Kuma instance polls our internal health endpoints every 60 to 300 seconds from outside our hosting provider — readiness, third-party dependencies, background queues, custom domains, and business-level checks such as payment credential validity. It runs on separate infrastructure, so an outage on our side does not take the watcher down with it.

Errors and scheduled jobs tracked in Sentry

Application errors are captured in Sentry, and every scheduled job checks in there when it starts and finishes — so a cron that silently stops running raises an alert rather than going unnoticed.

Alerts route to on-call with a written runbook

Alerts are graded P1 or P2 and pushed to dedicated on-call channels. Each alert type maps to a runbook with the same five sections: immediate containment, diagnosis, confirming recovery, escalation, and how to roll back or disable.

This page reports checks, not averages

Readings are taken at most every 30 seconds and cached for that long, and the page tells you exactly when the reading was taken. Where we can only verify that something is configured rather than live-probe it, the row says so in as many words.