Checkmate

Checkmate checks whether your servers and sites are reachable and performing, records what it finds, and tells you when that changes. Frontend and backend live in the same repository, and the project has been stress-tested past a thousand active monitors without hitting a bottleneck.

The lifecycle is deliberately dull. A check runs, its result is stored with a response time, and recent results are weighed against the monitor's status change threshold. Only when that threshold is crossed does state flip, an incident open or close, and notifications go out, which is what stops one flaky response from waking anyone.

The monitor types and outputs cover most of a small fleet.

• Check types: uptime, Docker, ping, SSL, port, game server, page speed and JSON query monitoring.
• Hardware metrics: memory, disk, CPU and network, through the separate Capture agent, with per-mountpoint disk selection.
• Status pages: public pages in four themes, built from the monitors you already have.
• Notifications: email, webhooks, Discord, Slack, PagerDuty, Matrix, Rocket.Chat, Microsoft Teams, Telegram, Pushover and Twilio SMS.
• Maintenance windows: scheduled work that does not raise incidents while it runs.

Deployment is a Docker Compose file starting the application image next to MongoDB, which is required and deliberately not embedded in it. It is light enough for a Raspberry Pi 4 or 5, expects a reverse proxy in front of it for TLS, and the interface is translated into sixteen languages.