nfs-provisioner

nfs-provisioner

nfs-provisioner is an out-of-tree dynamic provisioner for Kubernetes 1.4. You can use it to quickly & easily deploy shared storage that works almost anywhere. Or it can help you write your own out-of-tree dynamic provisioner by serving as an example implementation of the requirements detailed in the proposal.

It works just like in-tree dynamic provisioners: a StorageClass object can specify an instance of nfs-provisioner to be its provisioner like it specifies in-tree provisioners such as GCE or AWS. Then, the instance of nfs-provisioner will watch for PersistentVolumeClaims that ask for the StorageClass and automatically create NFS-backed PersistentVolumes for them. For more information on how dynamic provisioning works, see the docs or this blog post.

Quickstart

Choose some volume for your nfs-provisioner instance to store its state & data and mount it at /export in deploy/kube-config/deployment.yaml.

...
  volumeMounts:
    - name: export-volume
      mountPath: /export
volumes:
  - name: export-volume
    hostPath:
      path: /tmp/nfs-provisioner
...

Choose a provisioner name for a StorageClass to specify and set it in deploy/kube-config/deployment.yaml

...
args:
  - "-provisioner=example.com/nfs"
...

Create the deployment.

$ kubectl create -f deploy/kube-config/deployment.yaml
service "nfs-provisioner" created
deployment "nfs-provisioner" created

Create a StorageClass named "example-nfs" with provisioner: example.com/nfs.

$ kubectl create -f deploy/kube-config/class.yaml
storageclass "example-nfs" created

Create a PersistentVolumeClaim with annotation volume.beta.kubernetes.io/storage-class: "example-nfs"

$ kubectl create -f deploy/kube-config/claim.yaml
persistentvolumeclaim "nfs" created

A PersistentVolume is provisioned for the PersistentVolumeClaim. Now the claim can be consumed by some pod(s) and the backing NFS storage read from or written to.

$ kubectl get pv
NAME                                       CAPACITY   ACCESSMODES   RECLAIMPOLICY   STATUS      CLAIM         REASON    AGE
pvc-dce84888-7a9d-11e6-b1ee-5254001e0c1b   1Mi        RWX           Delete          Bound       default/nfs             23s

Deleting the PersistentVolumeClaim will cause the provisioner to delete the PersistentVolume and its data.

Deleting the provisioner deployment will cause any outstanding PersistentVolumes to become unusable for as long as the provisioner is gone.

Running

To deploy nfs-provisioner on a Kubernetes cluster see Deployment.

To use nfs-provisioner once it is deployed see Usage.

For information on running multiple instances of nfs-provisioner see Running Multiple Provisioners.

Implementation

The controller, the code for which is in the controller/ directory, watches PVCs and PVs to determine when to provision or delete volumes. It expects to receive an implementation of the Provisioner interface which has two methods: Provision and Delete. This NFS provisioner's implementation of the interface can be found under the volume/ directory.

So to create your own provisioner, you need to write your own implementation of the interface and pass it to the controller. Ideally you should be able to import the package to create the controller, without modifying any controller code. The passing in of the provisioner to the controller, and initialization of other things they might need (like a client for the Kubernetes API server), is done here in main.go.

Roadmap

This is still alpha/experimental and will change to reflect the out-of-tree dynamic provisioner proposal

November

• Create a process for releasing (to Docker Hub, etc.)
• Release 0.1 for kubernetes 1.5
• Support using the controller as a library
• Support running the provisioner as a StatefulSet

December

• Prevent multiple provisioners from racing to provision where possible (in a StatefulSet or DaemonSet)
• Add configurable retries for failed provisioning and deleting

Community, discussion, contribution, and support

Learn how to engage with the Kubernetes community on the community page.

You can reach the maintainers of this project at:

• Slack: #sig-storage

Kubernetes Incubator

This is a Kubernetes Incubator project. The project was established 2016-11-15. The incubator team for the project is:

• Sponsor: Clayton (@smarterclayton)
• Champion: Brad (@childsb)
• SIG: sig-storage

Code of conduct

Participation in the Kubernetes community is governed by the Kubernetes Code of Conduct.