K8S storage e2e experiment under VMware vSphere

Environment

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl version
Client Version: version.Info{Major:"1", Minor:"15+", GitVersion:"v1.15.3-beta.0.15+93d6c59069f682", GitCommit:"93d6c59069f6824fb05e926ea094966b47ed8b28", GitTreeState:"clean", BuildDate:"2019-08-15T03:43:36Z", GoVersion:"go1.12.5", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"15+", GitVersion:"v1.15.3-beta.0.15+93d6c59069f682", GitCommit:"93d6c59069f6824fb05e926ea094966b47ed8b28", GitTreeState:"clean", BuildDate:"2019-08-15T03:38:09Z", GoVersion:"go1.12.5", Compiler:"gc", Platform:"linux/amd64"}

And my K8S cluster is running on VMware PKS.

E2E steps

Create StorageClass

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: wcpglobal-storage-profile
parameters:
  storagePolicyID: eb806c5f-f214-4f34-9fa8-5387020354ea
provisioner: csi.vsphere.vmware.com
reclaimPolicy: Delete
volumeBindingMode: Immediate

You have to make sure your storagePolicyID has been created before from the vSphere UI, otherwise the provisioner won't work in terms of the dynamic provisioning

After you do kubectl apply to create the StorageClass, you will see the following:

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get storageclass wcpglobal-storage-profile -o yaml
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  creationTimestamp: "2019-10-01T20:56:02Z"
  name: wcpglobal-storage-profile
  resourceVersion: "7201419"
  selfLink: /apis/storage.k8s.io/v1/storageclasses/wcpglobal-storage-profile
  uid: 2ca6493e-b53b-40b5-bf56-340382e0dc84
parameters:
  storagePolicyID: eb806c5f-f214-4f34-9fa8-5387020354ea
provisioner: csi.vsphere.vmware.com
reclaimPolicy: Delete
volumeBindingMode: Immediate

Create PersistentVolumeClaim

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: demo-pvc
  namespace: my-podvm-ns
spec:
  storageClassName: wcpglobal-storage-profile
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 5Gi

After you do kubectl apply above yaml, you should be able to see the following details after a while(The status of your PVC might show Pending for a while):

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pvc -n my-podvm-ns
NAME       STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
demo-pvc   Bound    pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   15m
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pvc -n my-podvm-ns -o yaml
apiVersion: v1
items:
- apiVersion: v1
  kind: PersistentVolumeClaim
  metadata:
    annotations:
      kubectl.kubernetes.io/last-applied-configuration: |
        {"apiVersion":"v1","kind":"PersistentVolumeClaim","metadata":{"annotations":{},"name":"demo-pvc","namespace":"my-podvm-ns"},"spec":{"accessModes":["ReadWriteOnce"],"resources":{"requests":{"storage":"5Gi"}},"storageClassName":"wcpglobal-storage-profile"}}
      pv.kubernetes.io/bind-completed: "yes"
      pv.kubernetes.io/bound-by-controller: "yes"
      volume.beta.kubernetes.io/storage-provisioner: csi.vsphere.vmware.com
    creationTimestamp: "2019-10-02T21:59:58Z"
    finalizers:
    - kubernetes.io/pvc-protection
    name: demo-pvc
    namespace: my-podvm-ns
    resourceVersion: "7544761"
    selfLink: /api/v1/namespaces/my-podvm-ns/persistentvolumeclaims/demo-pvc
    uid: bbc40450-e8f9-4699-95dc-af3a09710b5f
  spec:
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 5Gi
    storageClassName: wcpglobal-storage-profile
    volumeMode: Filesystem
    volumeName: pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f
  status:
    accessModes:
    - ReadWriteOnce
    capacity:
      storage: 5Gi
    phase: Bound
kind: List
metadata:
  resourceVersion: ""
  selfLink: ""

Create Pod

apiVersion: v1
kind: Pod
metadata:
  name: demo-pod
  namespace: my-podvm-ns
spec:
  volumes:
    - name: demo-storage
      persistentVolumeClaim:
        claimName: demo-pvc
  containers:
    - name: demo-pod
      image: nginx
      ports:
        - containerPort: 80
          name: "http-server"
      volumeMounts:
        - mountPath: "/usr/share/nginx/html"
          name: demo-storage

After you apply above yaml. your pod will be under pending status for a while, that is because the dynamic provisioning is trying to provision a volume for you. It might take some time depends on the Storage Provider.

After a while, your pod should be up and running:

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pods -n my-podvm-ns
NAME       READY   STATUS    RESTARTS   AGE
demo-pod   1/1     Running   0          9m31s

You should see PV gets created automatically

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pv -n my-podvm-ns
NAME                                       CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS     CLAIM                                    STORAGECLASS                REASON   AGE
pvc-0c85c6fa-db94-4fbd-b8f5-7f3b8db85ca1   5Gi        RWO            Delete           Released   my-podvm-ns/demo-pvc                     wcpglobal-storage-profile            35m

What if the StorageClass gets deleted, what will happen to those Pods, PV, PVC ?

I have the following running now:

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pods,pvc,pv,storageclass -n my-podvm-ns
NAME           READY   STATUS    RESTARTS   AGE
pod/demo-pod   1/1     Running   0          18m

NAME                             STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
persistentvolumeclaim/demo-pvc   Bound    pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   21m

NAME                                                        CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS     CLAIM                                    STORAGECLASS                REASON   AGE
persistentvolume/pvc-1588c800-98ed-4a06-9d9d-48df8b010b75   1Gi        RWO            Delete           Bound      storage-policy-test/wcp-profile          wcp-profile-1jxtwc1tm3               22d
persistentvolume/pvc-76cada0f-6685-4684-a0ce-bda9e20bcbf2   1Gi        RWO            Delete           Bound      storage-policy-test/wcp-profile-second   wcp-profile-1jxtwc1tm3               22d
persistentvolume/pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            Delete           Bound      my-podvm-ns/demo-pvc                     wcpglobal-storage-profile            21m

NAME                                                                   PROVISIONER              AGE
storageclass.storage.k8s.io/0-wcp-test-storage-policyname-1jxtwc1tm3   csi.vsphere.vmware.com   22d
storageclass.storage.k8s.io/demo-storageclass                          csi.vsphere.vmware.com   26m
storageclass.storage.k8s.io/wcp-profile-1jxtwc1tm3                     csi.vsphere.vmware.com   46m
storageclass.storage.k8s.io/wcpglobal-storage-profile                  csi.vsphere.vmware.com   25h

Let me try to delete storage class, which is successfully deleted.

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl delete storageclass wcpglobal-storage-profile
storageclass.storage.k8s.io "wcpglobal-storage-profile" deleted
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get storageclass
NAME                                       PROVISIONER              AGE
0-wcp-test-storage-policyname-1jxtwc1tm3   csi.vsphere.vmware.com   22d
demo-storageclass                          csi.vsphere.vmware.com   27m
wcp-profile-1jxtwc1tm3                     csi.vsphere.vmware.com   47m

Surprisingly, the pod, pvc, pv are still there

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pods,pvc,pv -n my-podvm-ns
NAME           READY   STATUS    RESTARTS   AGE
pod/demo-pod   1/1     Running   0          21m

NAME                             STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
persistentvolumeclaim/demo-pvc   Bound    pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   24m

NAME                                                        CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS     CLAIM                                    STORAGECLASS                REASON   AGE
persistentvolume/pvc-1588c800-98ed-4a06-9d9d-48df8b010b75   1Gi        RWO            Delete           Bound      storage-policy-test/wcp-profile          wcp-profile-1jxtwc1tm3               22d
persistentvolume/pvc-76cada0f-6685-4684-a0ce-bda9e20bcbf2   1Gi        RWO            Delete           Bound      storage-policy-test/wcp-profile-second   wcp-profile-1jxtwc1tm3               22d
persistentvolume/pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            Delete           Bound      my-podvm-ns/demo-pvc                     wcpglobal-storage-profile            24m

kubectl exec into the demo pod, you will see you still have 5Gi capacity under the filesystem as you claimed from PVC:

root@demo-pod:/# df -h
Filesystem      Size  Used Avail Use% Mounted on
overlay         253M   14M  240M   6% /
/dev/sdb        4.9G   20M  4.9G   1% /usr/share/nginx/html
tmpfs           245M   12K  245M   1% /run/secrets/kubernetes.io/serviceaccount

Now, if you try to create another PVC by using the same storageclass, your PVC status will be under Pending.

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pvc -n my-podvm-ns
NAME         STATUS    VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
demo-pvc     Bound     pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   38m
demo-pvc-2   Pending                                                                        wcpglobal-storage-profile   3m47s

If you re-create the StorageClass you just deleted, the PVC status will be changed to Bound after a while.

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl apply -f wcpglobal-storage-profile-storageclass.yaml
storageclass.storage.k8s.io/wcpglobal-storage-profile created
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]#
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]#
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get storageclass
NAME                                       PROVISIONER              AGE
0-wcp-test-storage-policyname-1jxtwc1tm3   csi.vsphere.vmware.com   22d
demo-storageclass                          csi.vsphere.vmware.com   46m
wcp-profile-1jxtwc1tm3                     csi.vsphere.vmware.com   66m
wcpglobal-storage-profile                  csi.vsphere.vmware.com   6s
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pvc -n my-podvm-ns
NAME         STATUS    VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
demo-pvc     Bound     pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   42m
demo-pvc-2   Pending                                                                        wcpglobal-storage-profile   7m49s
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pvc -n my-podvm-ns
NAME         STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
demo-pvc     Bound    pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   44m
demo-pvc-2   Bound    pvc-0066d14e-8abf-4cab-aaa6-28b0ade16b66   5Gi        RWO            wcpglobal-storage-profile   9m21s

Conclusion: If you delete the storageclass, it will not affect your running Pod, PV, PVC. They will be still up and running. But you will not be able to create a new PVC by using the storageclass name you just deleted.

What will happen if I delete the PVC while a Pod is referencing to it ?

When I try to delete PVC, it seems get stuck(the terminal never returns):

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl delete -f pvc.yaml
persistentvolumeclaim "demo-pvc" deleted
^C
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pods,pvc -n my-podvm-ns
NAME           READY   STATUS    RESTARTS   AGE
pod/demo-pod   1/1     Running   0          69m

NAME                               STATUS        VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS                AGE
persistentvolumeclaim/demo-pvc     Terminating   pvc-bbc40450-e8f9-4699-95dc-af3a09710b5f   5Gi        RWO            wcpglobal-storage-profile   73m
persistentvolumeclaim/demo-pvc-2   Bound         pvc-0066d14e-8abf-4cab-aaa6-28b0ade16b66   5Gi        RWO            wcpglobal-storage-profile   38m

Now, I delete the Pod which is succeeded, and the PVC is deleted as well

root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl delete -f pod.yaml
pod "demo-pod" deleted
root@4201c9358d96fda3a30840a3e7cc9795 [ ~ ]# kubectl get pods,pvc -n my-podvm-ns

Conclusion: If a Pod is still using the PVC, the PVC could not be deleted, it will under terminating status forever until you have the Pod deleted.

PreviousVMware CNS NextExperiment on Persistent Volume Access Mode

Last updated 6 years ago

hashtagEnvironment

hashtagE2E steps

hashtagWhat if the StorageClass gets deleted, what will happen to those Pods, PV, PVC ?

hashtagConclusion: If you delete the storageclass, it will not affect your running Pod, PV, PVC. They will be still up and running. But you will not be able to create a new PVC by using the storageclass name you just deleted.

hashtagWhat will happen if I delete the PVC while a Pod is referencing to it ?

hashtagConclusion: If a Pod is still using the PVC, the PVC could not be deleted, it will under terminating status forever until you have the Pod deleted.