Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-37815

When node shutdown, the Pod whereabouts IP cannot be released (for a stateless application)

XMLWordPrintable

    • Moderate
    • No
    • CNF Network Sprint 257
    • 1
    • False
    • Hide

      None

      Show
      None
    • Release Note Not Required
    • In Progress

      This is a clone of issue OCPBUGS-37813. The following is the description of the original issue:

      This is a clone of issue OCPBUGS-37707. The following is the description of the original issue:

      This is a clone of issue OCPBUGS-29664. The following is the description of the original issue:

      Description of problem:

      Created Net-attach-def with 2 IPs in range. After that created deployment with 2 replicas using that net-attach-def. Whereabouts daemoneset is created also cronjob is enable reconsiling at every one min. 
      When i poweroff the node one which one of pod is deployded gracefully(poweroff)/ungracefully(poweroff --force) new pod is getting created on healthy node and stuck in container creating state

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

          4.14.11

      How reproducible:

      - Create whereabout daemon set with help of [documentation]([https://docs.openshift.com/container-platform/4.14/networking/multiple_networks/configuring-additional-network.html#nw-multus-creating-whereabouts-reconciler-daemon-set_configuring-additional-network)]
      - Update the reconciler_cron_expression to: "*/1 * * * *"
      - Create net-attach-def with 2 IPs in range
      - Create deployment with 2 replicas
      - Powreoff the node on which on of the POD is running
      - New Pod spawned on new healthy node with Container Creating in status.

      Steps to Reproduce:

      1. On fresh cluster with version 4.14.11
      2. Create whereabout daemon set with help of documentation   
      3. Update the reconciler_cron_expression to: "*/1 * * * *"
      $ oc create configmap whereabouts-config -n openshift-multus --from-literal=reconciler_cron_expression="*/1 * * * *"
      
      4. Create new project
      $ oc new-project nadtesting
      
      5. Apply below nad.yaml
      $ cat nad.yaml 
      apiVersion: "k8s.cni.cncf.io/v1"
      kind: NetworkAttachmentDefinition
      metadata:
        name: macvlan-net-attach1
      spec:
        config: '{
            "cniVersion": "0.3.1",
            "type": "macvlan",
            "master": "br-ex",
            "mode": "bridge",
            "ipam": {
              "type": "whereabouts",
              "datastore": "kubernetes",
              "range": "172.17.20.0/24",
              "range_start": "172.17.20.11",
              "range_end": "172.17.20.12"
            }
          }'
      
      6. Create deployment using net-attach-def with two replica,
      $ cat naddeployment.yaml 
      apiVersion: apps/v1
      kind: Deployment
      metadata:
        name: deployment1
        labels:
          app: macvlan1
      spec:
        replicas: 2
        selector:
          matchLabels:
            app: macvlan1
        template:
          metadata:
            annotations:
                 k8s.v1.cni.cncf.io/networks: macvlan-net-attach1
            labels:
              app: macvlan1
          spec:
            containers:
            - name: google
              image: gcr.io/google-samples/kubernetes-bootcamp:v1
              ports:
              - containerPort: 8080
      
      7. Two Pod will be created
      $ oc get pods -o wide
      NAME                          READY   STATUS    RESTARTS   AGE   IP            NODE                                       NOMINATED NODE   READINESS GATES
      deployment1-fbfdf5cbc-d6sgr   1/1     Running   0          15m   10.129.2.9    ci-ln-xvfy762-c1627-h7xzk-worker-0-qvzq2   <none>           <none>
      deployment1-fbfdf5cbc-njkpz   1/1     Running   0          15m   10.128.2.16   ci-ln-xvfy762-c1627-h7xzk-worker-0-8bdfh   <none>           <none>
      
      8. Power off the node using debug
      $ oc debug node/ci-ln-xvfy762-c1627-h7xzk-worker-0-8bdfh 
      # chroot /host
      # shutdown
      
      9. Wait for sometime new pod will created on healthy node which stuck in containercreating 
      $ oc get pod -o wide
      NAME                          READY   STATUS              RESTARTS   AGE     IP            NODE                                       NOMINATED NODE   READINESS GATES
      deployment1-fbfdf5cbc-6cb8d   0/1     ContainerCreating   0          9m53s   <none>        ci-ln-xvfy762-c1627-h7xzk-worker-0-blzlk   <none>           <none>
      deployment1-fbfdf5cbc-d6sgr   1/1     Running             0          28m     10.129.2.9    ci-ln-xvfy762-c1627-h7xzk-worker-0-qvzq2   <none>           <none>
      deployment1-fbfdf5cbc-njkpz   1/1     Terminating         0          28m     10.128.2.16   ci-ln-xvfy762-c1627-h7xzk-worker-0-8bdfh   <none>           <none>
      
      10. Node status just for reference,
      $ oc get nodes  
      NAME                                       STATUS     ROLES                  AGE   VERSION
      ci-ln-xvfy762-c1627-h7xzk-master-0         Ready      control-plane,master   59m   v1.27.10+28ed2d7
      ci-ln-xvfy762-c1627-h7xzk-master-1         Ready      control-plane,master   59m   v1.27.10+28ed2d7
      ci-ln-xvfy762-c1627-h7xzk-master-2         Ready      control-plane,master   58m   v1.27.10+28ed2d7
      ci-ln-xvfy762-c1627-h7xzk-worker-0-8bdfh   NotReady   worker                 43m   v1.27.10+28ed2d7
      ci-ln-xvfy762-c1627-h7xzk-worker-0-blzlk   Ready      worker                 43m   v1.27.10+28ed2d7
      ci-ln-xvfy762-c1627-h7xzk-worker-0-qvzq2   Ready      worker                 43m   v1.27.10+28ed2d
      
      

      Actual results:

      Shutdown node's pod stuck in terminating state and not releasing IP. New Pod is stuck in container creating status.

      Expected results:

      New Pod should start smoothly on new-node.    

      Additional info:

      - Just for information : If i follow manual approach the this issue will resolve for that i need to follow this step
      1. remove that termination IP from overlapping
      $ oc delete overlappingrangeipreservations.whereabouts.cni.cncf.io <IP>
      
      2. remove that termination IP from ippools.whereabouts.cni.cncf.io
      $ oc edit ippools.whereabouts.cni.cncf.io <IP Pool> 
      Remove that stale IP from list
      
      Also, the whereabouts-reconciler logs on the Terminating pod's node report:
      2024-02-19T10:48:00Z [debug] Added IP 172.17.20.12 for pod nadtesting/deployment1-fbfdf5cbc-njkpz
      2024-02-19T10:48:00Z [debug] the IP reservation: IP: 172.17.20.12 is reserved for pod: nadtesting/deployment1-fbfdf5cbc-njkpz
      2024-02-19T10:48:00Z [debug] pod reference nadtesting/deployment1-fbfdf5cbc-njkpz matches allocation; Allocation IP: 172.17.20.12; PodIPs: map[172.17.20.12:{}]
      2024-02-19T10:48:00Z [debug] no IP addresses to cleanup
      2024-02-19T10:48:00Z [verbose] reconciler success
      
      i.e. it fails to recognize the need to remove the allocation.

       

            rh-ee-marguerr Marcelo Guerrero Viveros
            openshift-crt-jira-prow OpenShift Prow Bot
            Weibin Liang Weibin Liang
            Carlos Goncalves
            Votes:
            0 Vote for this issue
            Watchers:
            6 Start watching this issue

              Created:
              Updated:
              Resolved: