[Answered ]-Add health check to Kafka consumer k8s pods

You could add a simple GET /health endpoint that returns a 200 status.

Or if you want something more dynamic to what the Kafka consumer is doing, catch exceptions in your consumer and flip a boolean variable that is inspected as part of your /health route to return 500 code upon any severe error. Kubernetes then will terminate any pods with non-200 HTTP status probe

First of all, ask yourself two questions:

• Q1: If Kafka does not work, how many APIs or features of your application can continue to work? In another word, how important is Kafka to your application?
• Q2: If Kafka does not work, can it be solved by a restart of the container?

If the answer of Q1 is "some" or "many", then you don’t need to do anything in your readiness probe. Unless the answer of Q1 is "very few" or "none", then you should report 500 in your readiness probe.

If the answer of Q2 is "yes", then you can report 500 in your liveness probe, otherwise just report 200 OK.

The reason behind them can be found here: https://danielw.cn/health-check-probes-in-k8s