Hi I'm looking into deploying flink natively on k8s, and I wanted if anyone else does it - and if so, what are the best practices to handle such deployment, in terms of connecting it to a ci cd pipeline, integrating custom flink cluster setting not via command line arguments (if it's even possible - the documentation does not mention such mechanism) etc. Thanks in advance!

Hi, of course, a lot of people use Kubernetes for their Flink deployments. Please look into the Flink Kubernetes Operator project: https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-main/

You can install the Flink Operator on Kubernetes. From there on you can deploy your applications with a manifest file: apiVersion: flink.apache.org/v1beta1 kind: FlinkDeployment 1. Build your java application in a Jar file, 2. Create Docker image based on: flink:2.0.0-scala_2.12-java21 3. Create your FlinkDeployment yaml 4. Deploy on kubernetes

Unfortunately I cannot use the operator on the k8s cluster that I work on. It’s only political, but I just can’t. I need to use k8s without the operator.

https://nightlies.apache.org/flink/flink-docs-release-1.20/docs/deployment/resource-providers/native_kubernetes/ Might help you out

I've done that in the past. It's definitely more complex than running the operator. State is a bit trickier when rolling out.