Health Check

HealthCheckServiceProvider registers /up -- what a load balancer, a Kubernetes liveness/readiness probe, or an uptime monitor expects to find. It's registered by default in a generated project.

curl http://localhost:8000/up

{"status": "ok", "checks": {"database": "ok"}}

A healthy response is a 200 with "status": "ok". If any check fails, the response is a 503 with "status": "error" -- the status code is what a load balancer actually acts on to decide whether to route traffic here, so nothing consuming this endpoint needs to parse the body just to know the answer.

What it checks

Currently just database connectivity -- a real SELECT 1 over the configured connection, not merely "is a URL configured" -- and only when Database is bound in the container (DatabaseServiceProvider registered). An app with no database reports {"status": "ok", "checks": {}}; there's nothing to check, so nothing is skipped silently or reported as failing.

Configuration

• HEALTH_CHECK_ENABLED -- default true. Set false to turn the endpoint off entirely, e.g. if you'd rather not expose it publicly and probe something internal-only instead.
• HEALTH_CHECK_PATH -- default /up. Change it if your infrastructure expects a different convention (/healthz is common on Kubernetes).

Setup

# main.py
from zeython import Application, HealthCheckServiceProvider

app = Application()
app.register(HealthCheckServiceProvider)

Register it early -- before DatabaseServiceProvider is fine either way, since providers' register() phase runs to completion before any boot() runs (see Architecture), so the database binding is guaranteed to exist by the time the health check route boots, regardless of registration order.