I am curious to know if beam dataflow job can be run with kubernetes. I can see lot of spring dataflow jobs run from kubernetes but not beam dataflow.

I can see one example like https://github.com/sanderploegsma/beam-scheduling-kubernetes/blob/master/kubernetes/cronjob.yml

But this doesn't explain how to pass args parameters like

args: ["--runner=DataflowRunner --project=$peoject --gcpTempLocation=$gcptemp"]   

Expanding more on this from https://streambench.wordpress.com/2018/06/07/set-up-the-direct-runner-for-beam/

I want to deploy this part on kubernetes.

beam_app_direct:
container_name: "beam_direct_app"
image: "beam_direct_app"
build:
    context: .
    dockerfile: ./Dockerfile-direct
environment:
- YOUR_ENV_PARAMETER=42
- ANOTHER_ENV_PARAMETER=abc
- ...
links:
- ...
# volume:
# - ./your-beam-app /usr/src/your-beam-app
command: "bash ./init.sh"

but I do not get any idea, how it can be deployed.

Updating more details.

My Cronjob.yaml file

apiVersion: batch/v1
kind: Job
metadata:
      name: "cronjob"
spec:
    template:
    spec:
    containers:
     - name: campaignjob
      image: cronjob
      build:
      context: .
      dockerfile: ./Dockerfile
      command: "bash ./init.sh"
  restartPolicy: Never

kubectl apply -f cronjob.yaml --validate=false

I am getting following error.

The Job "cronjob" is invalid: * spec.template.spec.containers: Required value * spec.template.spec.restartPolicy: Unsupported value: "Always": supported values: OnFailure, Never

Update: I am very surprised. I realised its a case of wrong YAML file only but even after 4 days there is not a comment. I even send this issue to Google team but they are asking me to use other technology.