Uploaded image for project: 'OpenShift Virtualization'
  1. OpenShift Virtualization
  2. CNV-44355

Eviction of Completed vmi pod triggers a live migration

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done-Errata
    • Icon: Critical Critical
    • CNV v4.17.0
    • None
    • CNV Virtualization
    • None
    • None

      Description of problem:

      
      

      Version-Release number of selected component (if applicable):

      v1.3.0-rc.1
      

      How reproducible:

      Always
      

      Steps to Reproduce:

          1. Deploy https://github.com/kubevirt/kubevirt/releases/tag/v1.3.0-rc.1 with OCP 4.17
          2. Deploy the following yaml file:
      ```
      ---
      apiVersion: kubevirt.io/v1
      kind: VirtualMachineInstance
      metadata:
        labels:
          special: vmi-masquerade-1
        name: vmi-masquerade-1
        annotations:
          descheduler.alpha.kubernetes.io/request-evict-only: ""
      spec:
        evictionStrategy: LiveMigrate
        domain:
          devices:
            disks:
            - disk:
                bus: virtio
              name: containerdisk
            - disk:
                bus: virtio
              name: cloudinitdisk
            interfaces:
            - masquerade: {}
              name: testmasquerade
              ports:
              - name: http
                port: 80
                protocol: TCP
            rng: {}
          resources:
            requests:
              memory: 1024M
        networks:
        - name: testmasquerade
          pod: {}
        terminationGracePeriodSeconds: 0
        volumes:
        - containerDisk:
            image: quay.io/kubevirt/fedora-with-test-tooling-container-disk:20240710_1265d1090
          name: containerdisk
        - cloudInitNoCloud:
            networkData: |
              version: 2
              ethernets:
                eth0:
                  addresses: [ fd10:0:2::2/120 ]
                  dhcp4: true
                  gateway6: fd10:0:2::1
            userData: |-
              #cloud-config
              password: fedora
              chpasswd: { expire: False }
              packages:
                - nginx
              runcmd:
                - [ "systemctl", "enable", "--now", "nginx" ]
          name: cloudinitdisk
      ```
         3. evict virt-launcher-vmi-masquerade-1-XXXXX pod and wait until a new pod virt-launcher-vmi-masquerade-1-YYYYY and the live migration is done. virt-launcher-vmi-masquerade-1-XXXXX pod is in the complete state.
         4. evict virt-launcher-vmi-masquerade-1-YYYYY pod and wait until a new pod virt-launcher-vmi-masquerade-1-ZZZZZ and the live migration is done. virt-launcher-vmi-masquerade-1-YYYYY pod is in the complete state.
         5. evict virt-launcher-vmi-masquerade-1-XXXXX pod, the pod gets deleted (sometimes, even this step triggers the live migration)
         6. evict virt-launcher-vmi-masquerade-1-YYYYY pod, live migration of virt-launcher-vmi-masquerade-1-ZZZZZ is started
      

      Actual results:

      virt-launcher-vmi-masquerade-1-ZZZZZ is live migrated after a completed virt-launcher-vmi-masquerade-1-YYYYY pod is evicted again
      

      Expected results:

      virt-launcher-vmi-masquerade-1-ZZZZZ is not live migrated after any completed virt-launcher-vmi-masquerade-1-????? pod is evicted
      

      Additional info:

      Running https://github.com/rajatjindal/kubectl-evict-pod for the pod eviction
      

              ffossemo@redhat.com Federico Fossemo
              jchaloup@redhat.com Jan Chaloupka
              Kedar Bidarkar Kedar Bidarkar
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

                Created:
                Updated:
                Resolved: