Monitoring
Health & Checks
Each scheduled probe stores a check result. Project health is computed only from those results — never from invented uptime.
Checks
A check is one execution of a monitor rule: timestamp, outcome (pass/fail), latency where applicable, and provider-specific evidence (HTTP status, certificate expiry, DNS answers). Continuous execution is driven by Trigger workers claiming due rules — see Monitoring overview.
Health
The health engine builds an evidence bag from rules → latest checks → results, then runs ordered contributors. Missing monitors or missing results produce honest “not configured” / “no data” outcomes instead of a fabricated score.
Contributors
- HTTP availability and HTTP response time — from HTTP monitors.
- SSL certificate — from SSL monitors.
- DNS resolution — from DNS / domain monitors.
Roadmap catalog kinds have future contributor stubs that do not score production health until the feature ships.
Empty states
No monitors configured, no successful checks yet, or Trigger not running all produce empty or partial health — that is intentional. Do not treat a blank health panel as “everything is fine.”
Limitations
- Health never invents availability from calendar time alone.
- Coming-soon monitor kinds do not contribute until activatable.
- Status page and incident impact use observed monitor state — wire those features separately.