Health Checks
On both Cloud Foundry and Kubernetes, it is possible to provide two separate endpoints for liveness checks ("are you alive?") and readiness checks ("are you ready for more requests?"). A failure on the former leads to a restart, whereas a failure on the latter temporarily takes the app instance out of the request dispatching rotation until a subsequent readiness probe is successful.
Learn more about health checks on Cloud Foundry.
Learn more about health checks on Kubernetes.
From @sap/cds^7.8 onwards, the Node.js runtime provides an out of the box endpoint for liveness and readiness checks at /health. Requests that reach this public endpoint are answered with the status code 200 and the body { status: 'UP' }.
For adding health checks to your Java app, please see Spring Boot Health Checks.
To achieve this, from @sap/cds-dk^7.8 onwards, the configuration for readiness health checks is included in the MTA template for deployment to CF. The default configuration specifies checks via http to / for Java (as there aren't any default Spring Boot health check endpoints) and /health for Node.js, respectively.
Additionally, for Node.js, the Helm chart template now specifies /health instead of / as the endpoint for the checks. For your app to be considered "alive", the endpoint /health needs to return a success response code. This means, if you have a fully custom server.js, you will need to add the /health endpoint to it, or adjust your MTA descriptor/Helm chart (if it's generated with @sap/cds-dk^7.8).
WARNING
Don't forget to adjust the values in your MTA descriptor/Helm chart in case you add the Spring Boot health check endpoints, or use a fully custom server.js, etc.!