I am afraid the current flink-kubernetes-operator always overrides the "REST_SERVICE_EXPOSED_TYPE" to "ClusterIP". Could you please share why the ingress[1] could not meet your requirements? Compared with NodePort, I think it is a more graceful implementation.
[1]. https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.1/docs/operations/ingress/ Best, Yang Vignesh Kumar Kathiresan via user <user@flink.apache.org> 于2022年9月2日周五 04:57写道： > Hello Flink community, > > Need some help with "flink kubernetes operator" based cluster setup. > > My flink cluster is set up using the flink-kubernetes-operator in AWS EKS. > The required resources(deployments, pods, services, configmaps etc) are > created as expected. But the service "*-rest" is created as a "ClusterIP" > type. I would want it created as a NodePort type. > > I want to expose the UI to external viewing via ingress using the aws alb > class. This aws-load balancer-controller requires my service to be of type > NodePort. > > I have tried a few options but the service is always created as ClusterIP. > 1) In the FlinkDeployment CRD, under spec.flinkConfiguration > added kubernetes.rest-service.exposed.type: "NodePort" > 2) In the operator helm values.yaml > > defaultConfiguration: > create: true > # Set append to false to replace configuration files > append: true > flink-conf.yaml: |+ > # Flink Config Overrides > kubernetes.rest-service.exposed.type: NodePort > > Neither option gives me a NodePort type service for the UI. > Any suggestions? > > > > > > > >
