Hello all. Just wanted to give a heads-up that Flink 1.20 images based on Java 17 have problems with launching jobs via Flink Kubernetes Operator. Those based on Java 11 work.
We are running a Flink Session cluster on Kubernetes, deploying it using Flink K8s Operator. Our session cluster was running custom Docker images, based on flink:1.20.2-java17 (we packed it with our jobs and libraries). When trying to deploy a Flink job using K8s Operator, it was failing with weird errors, basically complaining on the REST client failing. Didn’t say what exactly the problem was. When we introduced a separate deployment with all of our jobs, running in the same namespace, we could deploy jobs. We would log onto the “deployer” pod and run “flink run —python workflows/EQUIP/process.py —pyfiles workflows/lib/client_workflow-1.0.0-py3-none-any.whl”. That would deploy the job and it ran OK. When we switched to using flink:1.20.2 as the base image, we could deploy jobs using K8s Operator. I have no idea what the cause of this behavior could be, so this is a heads-up for all. If anyone has an idea, please chime in. Nix.