I have flink application deployed in k8s in standalone mode. Below are configurations set I have used in flink-config.yaml jobmanager.rpc.address: flink-jobmanager taskmanager.numberOfTaskSlots: 2 blob.server.port: 5124 jobmanager.rpc.port: 5123 taskmanager.rpc.port: 5122 queryable-state.proxy.ports: 5125 jobmanager.memory.process.size: 4092m jobmanager.memory.heap.size: 3092m taskmanager.memory.process.size: 3500m taskmanager.memory.task.heap.size: 1000m parallelism.default: 2 state.checkpoints.dir: gs://example web.upload.dir: /opt/flink/ state.backend: rocksdb state.checkpoints.dir: gs://example state.savepoints.dir: gs://example state.backend.incremental: true After submitting job with in few min JVM heap memory is filling up and getting restarted(I did enabled checkpoints). Job is running fine in my local but I am seeing this issue when submitting to k8s Any leads here will be very helpful. Thank you

Does your pod have enough resources? How have you deployed the application? Are you using Flink's native k8s CLI or a deployment/pod spec? Is the job manager or the task-manager dying of OOM?

our pods have 3GB memory allocated(out of 4 GB(process memory) we gave 3GB to heap), We deployed flink application as k8s standalone(session mode). our JM is dying of OOM