Kubernetes: Can I mount different subPaths from the same PV onto different locations of the same container?

Question

Can I mount different subPaths from the same PV onto different locations of the same container?

I run several wordpress instances on my company's Kubernetes cluster. Each instance has its own persistency volume and a container. The only peculiarity of my setup, is that I mount several paths of the PV onto several paths of the container.

All my containers worked well since a couple of weeks ago, when we upgraded Kubernetes to the current version. Since then, the hell began.

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"10", GitVersion:"v1.10.11", GitCommit:"637c7e288581ee40ab4ca210618a89a555b6e7e9", GitTreeState:"clean", BuildDate:"2018-11-26T14:38:32Z", GoVersion:"go1.9.3", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"10", GitVersion:"v1.10.11", GitCommit:"637c7e288581ee40ab4ca210618a89a555b6e7e9", GitTreeState:"clean", BuildDate:"2018-11-26T14:25:46Z", GoVersion:"go1.9.3", Compiler:"gc", Platform:"linux/amd64"}

When restarting a pod, if it gets scheduled to run on a different node, it get stuck on PodInitializing with the following event message

Multi-Attach error for volume "pvc-ac6b35f3-7716-11e8-adda-b60483de6a40" Volume is already exclusively attached to one node and can't be attached to another

Here are my resources.

---

 A Ceph RBD PersistentVolume

It contains two directories and a file

• html/: directory with php files
• logs/: directory with log files
• container-data.txt: a text file with some info

Defined as:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: rbd-wordpress-mysite
  labels:
    app: wordpress
spec:
  accessModes:
  - ReadWriteOnce
  resources:
    requests:
      storage: 5Gi

---

My pod

kind: Deployment
apiVersion: apps/v1beta1
metadata:
  name: wordpress-mysite
  labels:
    app: wordpress
  namespace: unibz
spec:
  template:
    metadata:
      name: wordpress-mysite
      labels:
        app: wordpress
      namespace: unibz
    spec:
      containers:
        - name: wordpress-mysite
          image: myimage
          volumeMounts:
          - mountPath: "/root/container-data.txt"
            name: wordpress-data
            subPath: container-data.txt
          - mountPath: "/var/www/html"
            name: wordpress-data
            subPath: html
          - mountPath: "/var/log/apache2"
            name: wordpress-data
            subPath: logs
          ports:
          - containerPort: 80
            name: wordpress-http
      volumes:
      - name: wordpress-data
        persistentVolumeClaim:
           claimName: rbd-wordpress-mysite
      - name: wordpress-conf
        configMap:
          name: wordpress-conf

---

Is this way of using the persistency wrong? Could it be the cause of the Multi-Attach error?

Answer 1

Looks like you are trying to attache the same PVC into a different node.

Access Modes Claims use the same conventions as volumes when requesting storage with specific access modes

In your .yaml I can see, you have setup accessModes: ReadWriteOnce.

A PersistentVolume can be mounted on a host in any way supported by the resource provider. As shown in the table below, providers will have different capabilities and each PV’s access modes are set to the specific modes supported by that particular volume. For example, NFS can support multiple read/write clients, but a specific NFS PV might be exported on the server as read-only. Each PV gets its own set of access modes describing that specific PV’s capabilities.

The access modes are:

• 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

From Kubernetes docs regarding Persistent Volumes you can read that CephFS does support all accessModes