Service status

Recent incidents

How this is measured

Two independent processes check each service every 60 seconds: the website itself, and the Discord bot. The bot matters because a site cannot report its own outage, so it watches the website from outside, keeps what it saw while the site was unreachable, and fills it in afterwards. Each bar is one hour and shows the worst state either observer recorded in it.

Operational means responding normally. Degraded means it is answering but something is off: slow responses, or leaderboard data that has stopped refreshing. Outage means it did not respond at all. A grey bar means nobody was watching for that hour, which is not the same as everything being fine, so it is never drawn green.

The Embark leaderboard row is not our service. It is the upstream board this whole site is built from, and we watch it because it can fail silently: the site stays up, the API answers, and every number simply stops changing. We check it three ways, all from the data we already collect every 10 minutes, without making a single extra request of Embark. We compare the board's own "last updated" stamp against the clock (normally under 25 minutes old), we watch whether any rank on the board is still moving at all (a normal hour has hundreds of changes and the quietest hour we have ever recorded had 112), and we record whether the fetch itself succeeded. If this row is red, our services are fine and the data is frozen, which is why an upstream problem never shows as a major outage in the banner above.

What this does not cover: both observers run on the same machine as the services, so this reports service-level failure, not a host or network failure. If the machine itself is down, this page will not load at all, and the history will show the gap once it returns.