Persistent Volume Claim Protection

FEATURE STATE: Kubernetes v1.9 alpha

This feature is currently in a alpha state, meaning:

The version names contain alpha (e.g. v1alpha1).
Might be buggy. Enabling the feature may expose bugs. Disabled by default.
Support for feature may be dropped at any time without notice.
The API may change in incompatible ways in a later software release without notice.
Recommended for use only in short-lived testing clusters, due to increased risk of bugs and lack of long-term support.

As of Kubernetes 1.9, persistent volume claims (PVCs) that are in active use by a pod can be protected from pre-mature removal.

Before you begin
PVC Protection Verification

Before you begin

A v1.9 or higher Kubernetes must be installed.
As PVC Protection is a Kubernetes v1.9 alpha feature it must be enabled:
1. Admission controller must be started with the PVC Protection plugin.
2. All Kubernetes components must be started with the PVCProtection alpha features enabled.

PVC Protection Verification

The example below uses a GCE PD StorageClass, however, similar steps can be performed for any volume type.

Create a StorageClass for convenient storage provisioning:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: slow
provisioner: kubernetes.io/gce-pd
parameters:
  type: pd-standard

Verification scenarios follow below.

Scenario 1: The PVC is not in active use by a pod

Create a PVC:

kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: slzc
spec:
  accessModes:
    - ReadWriteOnce
  storageClassName: slow
  resources:
    requests:
      storage: 3.7Gi

Check that the PVC has the finalizer kubernetes.io/pvc-protection set:

$ kubectl describe pvc slzc
Name:          slzc
Namespace:     default
StorageClass:  slow
Status:        Bound
Volume:        pvc-bee8c30a-d6a3-11e7-9af0-42010a800002
Labels:        <none>
Annotations:   pv.kubernetes.io/bind-completed=yes
             pv.kubernetes.io/bound-by-controller=yes
             volume.beta.kubernetes.io/storage-provisioner=kubernetes.io/gce-pd
Finalizers:    [kubernetes.io/pvc-protection]
Capacity:      4Gi
Access Modes:  RWO
Events:
Type    Reason                 Age   From                         Message
----    ------                 ----  ----                         -------
Normal  ProvisioningSucceeded  2m    persistentvolume-controller  Successfully provisioned volume pvc-bee8c30a-d6a3-11e7-9af0-42010a800002 using kubernetes.io/gce-pd

Delete the PVC and check that the PVC (not in active use by a pod) was removed successfully.

Scenario 2: The PVC is in active use by a pod

Again, create the same PVC.
Create a pod that uses the PVC:

kind: Pod
apiVersion: v1
metadata:
  name: app1
spec:
  containers:
  - name: test-pod
    image: k8s.gcr.io/busybox:1.24
    command:
      - "/bin/sh"
    args:
      - "-c"
      - "date > /mnt/app1.txt; sleep 60 && exit 0 || exit 1"
    volumeMounts:
      - name: path-pvc
        mountPath: "/mnt"
  restartPolicy: "Never"
  volumes:
    - name: path-pvc
      persistentVolumeClaim:
        claimName: slzc

Wait until the pod status is Running, i.e. the PVC becomes in active use.
Delete the PVC that is now in active use by a pod and verify that the PVC is not removed but its status is Terminating:

Name:          slzc
Namespace:     default
StorageClass:  slow
Status:        Terminating (since Fri, 01 Dec 2017 14:47:55 +0000)
Volume:        pvc-803a1f4d-d6a6-11e7-9af0-42010a800002
Labels:        <none>
Annotations:   pv.kubernetes.io/bind-completed=yes
               pv.kubernetes.io/bound-by-controller=yes
               volume.beta.kubernetes.io/storage-provisioner=kubernetes.io/gce-pd
Finalizers:    [kubernetes.io/pvc-protection]
Capacity:      4Gi
Access Modes:  RWO
Events:
  Type    Reason                 Age   From                         Message
  ----    ------                 ----  ----                         -------
  Normal  ProvisioningSucceeded  52s   persistentvolume-controller  Successfully provisioned volume pvc-803a1f4d-d6a6-11e7-9af0-42010a800002 using kubernetes.io/gce-pd

Wait until the pod status is Terminated (either delete the pod or wait until it finishes). Afterwards, check that the PVC is removed.

Scenario 3: A pod starts using a PVC that is in Terminating state

Again, create the same PVC.
Create a first pod that uses the PVC:

kind: Pod
apiVersion: v1
metadata:
  name: app1
spec:
  containers:
  - name: test-pod
    image: k8s.gcr.io/busybox:1.24
    command:
      - "/bin/sh"
    args:
      - "-c"
      - "date > /mnt/app1.txt; sleep 600 && exit 0 || exit 1"
    volumeMounts:
      - name: path-pvc
        mountPath: "/mnt"
  restartPolicy: "Never"
  volumes:
    - name: path-pvc
      persistentVolumeClaim:
        claimName: slzc

Wait until the pod status is Running, i.e. the PVC becomes in active use.
Delete the PVC that is now in active use by a pod and verify that the PVC is not removed but its status is Terminating:

Name:          slzc
Namespace:     default
StorageClass:  slow
Status:        Terminating (since Fri, 01 Dec 2017 14:47:55 +0000)
Volume:        pvc-803a1f4d-d6a6-11e7-9af0-42010a800002
Labels:        <none>
Annotations:   pv.kubernetes.io/bind-completed=yes
               pv.kubernetes.io/bound-by-controller=yes
               volume.beta.kubernetes.io/storage-provisioner=kubernetes.io/gce-pd
Finalizers:    [kubernetes.io/pvc-protection]
Capacity:      4Gi
Access Modes:  RWO
Events:
  Type    Reason                 Age   From                         Message
  ----    ------                 ----  ----                         -------
  Normal  ProvisioningSucceeded  52s   persistentvolume-controller  Successfully provisioned volume pvc-803a1f4d-d6a6-11e7-9af0-42010a800002 using kubernetes.io/gce-pd

Create a second pod that uses the same PVC:

kind: Pod
apiVersion: v1
metadata:
  name: app2
spec:
  containers:
  - name: test-pod
    image: gcr.io/google_containers/busybox:1.24
    command:
      - "/bin/sh"
    args:
      - "-c"
      - "date > /mnt/app1.txt; sleep 600 && exit 0 || exit 1"
    volumeMounts:
      - name: path-pvc
        mountPath: "/mnt"
  restartPolicy: "Never"
  volumes:
    - name: path-pvc
      persistentVolumeClaim:
        claimName: slzc

Verify that the scheduling of the second pod fails with the below warning:

Warning  FailedScheduling  18s (x4 over 21s)  default-scheduler  persistentvolumeclaim "slzc" is being deleted

Wait until the pod status of both pods is Terminated (either delete the pods or wait until they finish). Afterwards, check that the PVC is removed.

Create an Issue Edit this Page

Tasks

Step-by-step instructions for performing operations with Kubernetes.