Network & SecurityLoad Balancer

Health checks

The load balancer actively probes every backend on a fixed interval. A backend that fails enough consecutive probes is removed from rotation; it returns automatically once it passes again. Health checks are configured in the Routing & Health section of the create form.

Check types

TypeWhat it verifies
http (default)The backend returns a successful response for a GET on the configured Path
httpsSame as http, over TLS
tcpA TCP connection can be established on the backend port — no path involved

Use http/https whenever your service exposes a health endpoint — a TCP check only proves the port is open, not that the application can serve requests.

Settings

FieldDefaultDescription
Path/healthRequest path for http/https checks
Interval10000 msTime between probes
Timeout2000 msHow long a probe waits before counting as a failure
Healthy threshold2Consecutive passes before a backend re-enters rotation
Unhealthy threshold3Consecutive failures before a backend is removed

With the defaults, a failing backend is removed after roughly 30 seconds (3 failures × 10 s interval) and restored about 20 seconds after it recovers.

Choosing a health endpoint

  • Return 200 only when the service can actually do work (dependencies reachable, not just process-up).
  • Keep the endpoint fast and unauthenticated — it’s probed every few seconds.
  • Don’t reuse a heavy endpoint; a dedicated /health route avoids cache and auth side effects.