Choose a check type
| Type | Use it to detect | Health signal |
|---|---|---|
| Ping / heartbeat | A scheduled job stops completing, or a reported value crosses a threshold | Your sender checks in |
| Website | An HTTP endpoint fails or returns the wrong status code | Pingstack performs a GET |
| DNS | Public DNS records disappear or differ from expectations | Answers from three public resolvers |
| TCP | A public service no longer accepts TCP connections | A connection to a hostname and port |
| Domain | A registration approaches expiration | An expiration date from a domain lookup |
| SSL certificate | A presented certificate approaches expiration | Certificate expiration over TLS |
A TCP success does not prove an application query works. A website success does not prove a background job ran. Use several checks when a service has independent failure modes.
Give checks names that identify both the service and the signal. Check kind cannot be changed after creation; create a replacement check when changing the monitoring method.
Website, DNS and TCP intervals are constrained by your plan. Domain and certificate lookups run hourly. Heartbeats follow the sender’s expected interval or cron schedule. Compare scheduling behavior.