These are in a service's Settings → General. Change the values, then click Save to apply them together.
- Replicas — how many copies of the app run. More than one keeps the app up while one restarts.
- Resource limits — the most CPU and memory one copy may use.
- Autoscaling — adds and removes copies between a minimum and a maximum as CPU use crosses a target. It needs a CPU limit to measure against.
- Health check — a path (such as
/health) wdot calls to know the app is ready. A copy that fails it receives no traffic until it recovers. - Rate limits — the number of requests one visitor may make per second, with a burst allowance. Requests over the limit are refused.