K
Q

GKE: right way to mount same PV on multiple pods

7/5/2017

I'm having trouble to create a persistent volume that I can use from different pods (1 write, another read).

Tried to use gcePersistentDisk directly in the pod spec like in the example on the k8s page (plus readOnly):

apiVersion: v1
kind: Pod
metadata:
  name: test-pd
spec:
  containers:
  - image: gcr.io/google_containers/test-webserver
    name: test-container
    volumeMounts:
    - mountPath: /test-pd
      name: test-volume
      readOnly: true
  volumes:
  - name: test-volume
    gcePersistentDisk:
      pdName: my-data-disk
      fsType: ext4
      readOnly: true

Then in the second pod spec exactly the same except the readOnly ... but got an NoDiskConflict error.

Second approach is to use PersistentVolume and PersistentVolumeClaim like this:

apiVersion: v1
kind: PersistentVolume
metadata:
  name: data-standard
spec:
  capacity:
    storage: 1Gi
  accessModes:
    - ReadWriteMany
  gcePersistentDisk:
    fsType: ext4
    pdName: data

---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: data-standard-claim
spec:
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 1Gi

But now I get an error telling me:

MountVolume.MountDevice failed for volume "kubernetes.io/gce-pd/xxx" (spec.Name: "yyy") pod "6ae34476-6197-11e7-9da5-42010a840186" (UID: "6ae34476-6197-11e7-9da5-42010a840186") with: mount failed: exit status 32 Mounting command: mount Mounting arguments: /dev/disk/by-id/google-gke-cluster-xxx /var/lib/kubelet/plugins/kubernetes.io/gce-pd/mounts/gke-cluster-xxx [ro] Output: mount: wrong fs type, bad option, bad superblock on /dev/sdb, missing codepage or helper program, or other error In some cases useful info is found in syslog - try dmesg | tail or so.
Error syncing pod, skipping: timeout expired waiting for volumes to attach/mount for pod "default"/"my-deployment". list of unattached/unmounted volumes=[data]

So what is the correct way of using a GCE disk with multiple pods.

PS: Kubernetes 1.6.6

-- pkyeck
google-kubernetes-engine
kubernetes

Similar Questions

Kubernetes client for Go - ERRORS only encoded map or array can be decoded into a struct
Kubernetes StatefulSets after cluster restart
Kubernetes application running on master - DaemonSet
Would Docker swarm assist me with managing containers with specific environment variables?

3 Answers

2/12/2019

i guess, you should be able to do this by using PV and PVC. let's take an example where you have one pv and PVC. i am not much sure on multi worker-node architecture. i am a minikube user. have a check here - https://github.com/kubernetes/kubernetes/issues/60903

Now you have PV and PVC, both status is bound. now you will use the claim in the pod/deployment resource definition and in pvc name you will use the same claim name. this will attach the volume to both the pods.

---
#pv defination:

---
kind: PersistentVolume
apiVersion: v1
metadata:
  name: pv0
spec:
  storageClassName: standard
  accessModes:
    - ReadWriteOnce
  capacity:
    storage: 5Gi
  hostPath:
    path: /path_on_node
    type: DirectoryOrCreate
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: claim0
spec:
  storageClassName: standard
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 5Gi
  volumeName: pv0
--- 
# pod1
          volumeMounts:
            - mountPath: /container_path
              name: vol0
              subPath: sub_path_on_pv(say pod1 so on disk data will be written at /path_on_node/pod1
      volumes:
        - name: vol0
          persistentVolumeClaim:
            claimName: claim0  
---
# pod2
          volumeMounts:
            - mountPath: /container_path
              name: vol2
              subPath: sub_path_on_pv(say pod2 so on disk data will be written at /path_on_node/pod2
      volumes:
        - name: vol2
          persistentVolumeClaim:
            claimName: claim0
-- Sandeep kumar singh
Source: StackOverflow
7/6/2017

According to https://kubernetes.io/docs/concepts/storage/persistent-volumes/#access-modes, GCE Disks do not support ReadWriteMany. I am not sure if this explains the issue but I would advise you to try another compatible Volume type.

-- Javier Salmeron
Source: StackOverflow
9/13/2017

Instead of ReadWriteMany, can you use ReadOnlyMany?

Accessmodes:

ReadWriteOnce – the volume can be mounted as read-write by a single node

ReadOnlyMany – the volume can be mounted read-only by many nodes

ReadWriteMany – the volume can be mounted as read-write by many nodes

GCE persistent disk do not support ReadWriteMany

Here is the list of providers and supported accessmodes:

https://kubernetes.io/docs/concepts/storage/persistent-volumes/#access-modes

-- Balaji Boggaram Ramanarayan
Source: StackOverflow