如何在 GKE 中使用 ReadWriteMany 创建持久卷声明？

Question

What is the best way to create a persistent volume claim with ReadWriteMany attaching the volume to multiple pods?

Based off the support table in https://kube.netes.io/docs/concepts/storage/persistent-volumes , GCEPersistentDisk does not support ReadWriteMany natively.

What is the best approach when working in the GCP GKE world? Should I be using a clustered file system such as CephFS or Glusterfs? Are there recommendations on what I should be using that is production ready?

I was able to get an NFS pod deployment configured following the steps here - https://medium.com/platformer-blog/nfs-persistent-volumes-with-kube.netes-a-case-study-ce1ed6e2c266 however it seems a bit hacky and adds another layer of complexity. It also seems to only allow one replica (which makes sense as the disk can't be mounted multiple times) so if/when the pod goes down, my persistent storage will as well.

Answer 1

I agree that it's disappointing but it's a consequence of the use of persistent disk which does not permit attaching to multiple instances read-write.

I've had success with NFS and with the limitations you describe.

You could -- as you state -- use Gluster or similar too.

A more expensive albeit managed Google Cloud alternative is Cloud Filestore: https://cloud.google.com/filestore/docs/accessing-fileshares

Your questions suggests that you need NFS-like semantics but, if you don't, you may consider using Google Cloud Storage.

Answer 2

It's possible now with Cloud Filestore .

First create a Filestore instance.

gcloud filestore instances create nfs-server
    --project=[PROJECT_ID]
    --zone=us-central1-c
    --tier=STANDARD
    --file-share=name="vol1",capacity=1TB
    --network=name="default",reserved-ip-range="10.0.0.0/29"

Then create a persistent volume in GKE.

apiVersion: v1
kind: PersistentVolume
metadata:
  name: fileserver
spec:
  capacity:
    storage: 1T
  accessModes:
  - ReadWriteMany
  nfs:
    path: /vol1
    server: [IP_ADDRESS]

[IP_ADDRESS] is available in filestore instance details.

You can now request a persistent volume claim.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: fileserver-claim
spec:
  accessModes:
  - ReadWriteMany
  storageClassName: "fileserver"
  resources:
    requests:
      storage: 100G

Finally, mount the volume in your pod.

apiVersion: v1
kind: Pod
metadata:
  name: my-pod
spec:
  containers:
  - name: my container
    image: nginx:latest
    volumeMounts:
    - mountPath: /workdir
      name: mypvc
  volumes:
  - name: mypvc
    persistentVolumeClaim:
      claimName: fileserver-claim
      readOnly: false

Solution is detailed here : https://cloud.google.com/filestore/docs/accessing-fileshares

Answer 3

Just in case you are using Terraform already to manage your GKE, you can also use this terraform module to handle "nfs-server" creation & management for you.

Personally, I found this very handy in my situation as it's cost-effective and I don't have to manually create the "nfs-server".