Hi. I have a request for

flink-kubernetes-operator

and am not sure where I shall put this request. The request is that

flink-kubernetes-operator

can support to set cpu limit and cpu request of job manager separately. In our Flink app, the cpu usage is pretty low at most of the time. However, when creating checkpoints & savepoints, the flink job manager gets cpu throttled due to heavy loading. If the

flink-kubernetes-operator

can allow us to set cpu limit and request separately, it will be beneficial for saving costs without performance degradation. Any comment is appreciated!

I guess, under pod template you can define resource limits and request

For native K8s integration(default mode in Flink K8s Operator), you could simply use

kubernetes.jobmanager.cpu.limit-factor

to control the resource limit. The resources limit cpu will be set to

cpu * limit-factor

.

Thank you all for the comments. For simplicity, I set the cpu limit via

kubernetes.jobmanager.cpu.limit-factor

suggested by Yang and it does work. Many thanks!

Seems

kubernetes.jobmanager.cpu.limit-factor

was added in flink >= 1.15.X. I am using 1.13.6 currently and tried following options. But din work.

taskManager:
    resource:
      memory: "2000m"
      cpu: "1"
    podTemplate:
      apiVersion: v1
      kind: Pod
      metadata:
        name: task-manager-pod-template
      spec:
        containers:
        - name: flink-main-container
          resources:
            requests:
              memory: "2000m"
              cpu: "1"
            limits:
              memory: "3000m"
              cpu: "2"

Section resource overrides the resources section inside containers section. Tried removing resources all together and that was setting CPU requests and limits equal to number of task slots . Its completely ignoring the resources section inside containers section

taskManager:
    podTemplate:
      apiVersion: v1
      kind: Pod
      metadata:
        name: task-manager-pod-template
      spec:
        containers:
        - name: flink-main-container
          resources:
            requests:
              memory: "2000m"
              cpu: "1"
            limits:
              memory: "3000m"
              cpu: "2"

Any options in flink 1.13.6 or does flink-kuberentes-operator provide any such options