Monitors

A monitor belongs to one component and moves it between operational and major outage on its own. You can always override a component by hand. Monitors check only what you configure, never faster than once a minute, and keep only pass or fail, the status code and the response time for 48 hours. See the bot page for how our checks identify themselves.

HTTP monitors

• Method GET or HEAD against an http or https URL on port 80, 443, 8080 or 8443.
• Passes when the status code is in your expected list (default 200-399; write 200,204,301-302 for a list).
• Timeout 1 to 10 seconds. Up to 3 redirects are followed and at most 64 KB of the response is read. The body is discarded.
• The interval is at least your plan's fastest check (see pricing) and never under 60 seconds.
• The component goes down after 2 failed checks in a row, which avoids flapping on a single bad response.
• Only public internet addresses can be monitored. Addresses on private networks, loopback, link-local and cloud metadata ranges are refused when you save and again at check time.

Heartbeat monitors

For cron jobs and anything that cannot be reached from outside. The job calls a URL when it runs; if no call arrives within the interval plus a grace period, the component goes down. The next call brings it back.

curl -fsS -m 10 https://status.starhold.app/hb/YOUR_HEARTBEAT_TOKEN

• Both GET and POST work and answer 204 with no body.
• The token is shown once, when you create the monitor. If you lose it, delete the monitor and make a new one.
• At most 60 calls a minute per token; more answer 429.