ScaledJob: Support metadata annotations for ephemeral volume claims

[dpertin]

Add x-kubernetes-preserve-unknown-fields to ScaledJob CRD in order to support metadata in ephemeral volume claims

Based on #1311

Previously, ArgoCD used to see a diff, bypassed the error but kept the application "OutOfSync":

Yet a warning was triggered: unknown field "spec.jobTargetRef.template.spec.volumes[5].ephemeral.volumeClaimTemplate.metadata.annotations"

Now the change is applied, the desired block is part of the "live manifest" and the application is "Synced":

Checklist

• I have verified that my change is according to the deprecations & breaking changes policy
• Changelog has been updated and is aligned with our changelog requirements
• Commits are signed with Developer Certificate of Origin (DCO - learn more)

Fixes #6254

[JorTurFer]

Thanks for the fix! He have merged this PR, and I'd say that it will fix the current issue too. Could you double check it?

[zroubalik]

@dpertin any update please?

[malcolm061990]

Hi @JorTurFer @zroubalik
The problem still exists deploying keda scaledjob using 2.16.1 chart:

        - ephemeral:
            volumeClaimTemplate:
              metadata:
                annotations:
                  annotaion: annotation
                labels:
                  label: label
              spec:
                accessModes:
                - ReadWriteOnce
                resources:
                  requests:
                    storage: 100Gi
          name: ephemeral-volume

kubectl apply --dry-run=server -f scaledjoboutput.yaml
Error from server (BadRequest): error when creating "scaledjoboutput.yaml": ScaledJob in version "v1alpha1" cannot be handled as a ScaledJob: strict decoding error: unknown field "spec.jobTargetRef.template.spec.volumes[2].ephemeral.volumeClaimTemplate.metadata.annotations", unknown field "spec.jobTargetRef.template.spec.volumes[2].ephemeral.volumeClaimTemplate.metadata.labels"

[JorTurFer]

Could you share a demo ScaledJob to test it on my local?

[malcolm061990]

Could you share a demo ScaledJob to test it on my local?

@JorTurFer
Sure, here it is:

---
apiVersion: keda.sh/v1alpha1
kind: ScaledJob
metadata:
  labels:
    name: godummy
  name: godummy
  namespace: default
spec:
  failedJobsHistoryLimit: 2
  jobTargetRef:
    activeDeadlineSeconds: 7200
    backoffLimit: 1
    completions: 1
    parallelism: 1
    template:
      metadata:
        annotations: null
      spec:
        containers:
        - command:
          - -c
          - job
          env: null
          image: godummy:main
          imagePullPolicy: Always
          name: godummy-container
          volumeMounts:
          - mountPath: /app/data/
            name: ephemeral-volume
          - mountPath: /tmp
            name: tmp
          - mountPath: /etc/ssl/certs/app-ca.pem
            name: vol-app-ca
            readOnly: true
            subPath: app-ca.pem
        imagePullSecrets:
        - name: registry-secret
        restartPolicy: Never
        volumes:
        - emptyDir: {}
          name: tmp
        - name: vol-app-ca
          secret:
            secretName: app-ca
        - ephemeral:
            volumeClaimTemplate:
              metadata:
                annotations:
                  annotaion: annotaion
                labels:
                  label: label
              spec:
                accessModes:
                - ReadWriteOnce
                resources:
                  requests:
                    storage: 100Gi
          name: ephemeral-volume
  maxReplicaCount: 1
  minReplicaCount: 0
  pollingInterval: 30
  successfulJobsHistoryLimit: 3
  triggers:
  - authenticationRef:
      kind: ClusterTriggerAuthentication
      name: trigger
    metadata:
      bootstrapServers: bootstrapServer1:9092,bootstrapServer2:9092,bootstrapServer3:9092
      consumerGroup: infra
      lagThreshold: "10000"
      scaleToZeroOnInvalidOffset: "true"
      topic: godummy-job
    type: kafka

[JorTurFer]

I've tested and the CRD in KEDA repo is correct and it keeps the annotation

I'm going to check using helm chart v2.16.1

[JorTurFer]

Helm chart doesn't work, there are missing changes related with that CRD in the chart. The problem is solved in KEDA repo, but helm chart is still not working. @zroubalik @wozniakjan , we should ensure that the CRDs are updated during the helm release process, probably I made some mistake during last release and the CRD isn't correctly placed there

I think that this PR isn't required as this repo is fixed

[malcolm061990]

thanks @JorTurFer for your help
please notify me in this ticket when it will be fixed :)