Shared Filesystem

Shared filesystems in the 1.5.4 release are currently in Technology Preview, and are only available using the CSI driver. This experimental feature is not yet intended for production use.

Shared Filesystem support allows volumes to be mounted for read & write access by multiple containers simultaneously, even from different nodes. In Kubernetes, shared filesystems are referred to as ReadWriteMany or RWX volumes.

See the architecture for more information about the Shared Filesystem components.

How to provision ReadWriteMany Volumes

RWX Volumes are dynamically provisioned by Ondat when using a Ondat Kubernetes StorageClass. To create an RWX volume set the AccessMode for the PVC to ReadWriteMany.

An example of a RWX PersistentVolumeClaim is shown below:

kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: shared-vol
  labels:
    storageos.com/replicas: "1" # Enable features using Labels
                                # You can enable replication in the
                                # StorageClass params to set defaults
spec:
  storageClassName: "fast" # Ondat StorageClass
  accessModes:
    - ReadWriteMany # AccessMode that triggers creation of an NFS based Ondat Volume
  resources:
    requests:
      storage: 20Gi

The following Deployment example shows how multiple Nginx Pods can mount the same shared filesystem PVC.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx
  labels:
    app: nginx
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.7.9
        ports:
        - containerPort: 80
        volumeMounts:
        - name: docroot
          mountPath: /usr/share/nginx/html
      volumes:
      - name: docroot
        persistentVolumeClaim:
          claimName: shared-vol

Only installations of Ondat using CSI support RWX Volumes.

Volume Deletion

Shared volumes can be deleted through standard Kubernetes PVC deletion. When a RWX PVC is deleted, Ondat acts differently according to the ReclaimPolicy of the PVC. For more information about the ReclaimPolicy field see the Kubernetes Documentation.

Reclaim Policy

The default reclaim policy for a PVC, PV pair is defined in the StorageClass. All PVs created by a StorageClass inherit the reclaimPolicy set in the StorageClass.

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  labels:
    app: storageos
  name: storageos
parameters:
  csi.storage.k8s.io/fstype: ext4
  pool: default
provisioner: storageos
reclaimPolicy: Delete # <----- Default reclaim policy

• Delete (default)

  If the RWX PVC is deleted then the RWX PV is deleted along with the NFS server StatefulSet and its underlying RWO volume. As such data written to the RWX volume is non-recoverable after deletion.

• Retain

  The RWX PVC is deleted but the PV remains available. The NFS server StatefulSet remains untouched along with its associated RWO PVC. Hence, it is possible to reuse the volume and have access to data written to the RWX volume after the RWX PVC has been deleted.

  In order to reuse the NFS server to serve a different RWX PVC, it is is necessary to “unlock” the PV, which only Kubernetes administrators with privileges can do. The PV is “unlocked” by deleteing the spec.claimRef field from the PV.

  For instance given a RWX PVC,pvc-1 was bound to the PV pv-1, before pvc-1 was deleted. Then in order to reuse pv-1 edit the PV, i.e kubectl edit pv pv-1 and delete the spec.claimRef attribute.

  Deletion of the spec.claimRef field makes the PV available for any new PVC whose requirements are met by the PV. Therefore creating a new RWX volume with the previously used; StorageClass, capacity, access type and filesystem will make the PVC bind to the previously used RWX PV, even if the new PVC is provisioned in a different namespace.