Driver for CIFS (SMB, Samba, Windows Share) network filesystems as Kubernetes volumes.
Docker containers running in Kubernetes have an ephemeral file system: Once a container is terminated, all files are gone. In order to store persistent data in Kubernetes, you need to mount a Persistent Volume into your container. Kubernetes has built-in support for network filesystems found in the most common cloud providers, like Amazon's EBS, Microsoft's Azure disk, etc. However, some cloud hosting services, like the Hetzner cloud, provide network storage using the CIFS (SMB, Samba, Windows Share) protocol, which is not natively supported in Kubernetes.
Fortunately, Kubernetes provides Flexvolume, which is a plugin mechanism enabling users to write their own drivers. There are a few flexvolume drivers for CIFS out there, but for different reasons none of them seemed to work for me. So I wrote my own, which can be found on github.com/fstab/cifs.
The flexvolume plugin is a single shell script named cifs. This shell script must be available on the Kubernetes master and on each of the Kubernetes nodes. By default, Kubernetes searches for third party volume plugins in /usr/libexec/kubernetes/kubelet-plugins/volume/exec/
. The plugin directory can be configured with the kubelet's --volume-plugin-dir
parameter, run ps aux | grep kubelet
to learn the location of the plugin directory on your system (see #1). The cifs
script must be located in a subdirectory named fstab~cifs/
. The directory name fstab~cifs/
will be mapped to the Flexvolume driver name fstab/cifs
.
On the Kubernetes master and on each Kubernetes node run the following commands:
VOLUME_PLUGIN_DIR="/usr/libexec/kubernetes/kubelet-plugins/volume/exec"
mkdir -p "$VOLUME_PLUGIN_DIR/fstab~cifs"
cd "$VOLUME_PLUGIN_DIR/fstab~cifs"
curl -L -O https://raw.githubusercontent.com/fstab/cifs/master/cifs
chmod 755 cifs
The cifs
script requires a few executables to be available on each host system:
mount.cifs
, on Ubuntu this is in the cifs-utils package.jq
, on Ubuntu this is in the jq package.mountpoint
, on Ubuntu this is in the util-linux package.base64
, on Ubuntu this is in the coreutils package.To check if the installation was successful, run the following command:
VOLUME_PLUGIN_DIR="/usr/libexec/kubernetes/kubelet-plugins/volume/exec"
$VOLUME_PLUGIN_DIR/fstab~cifs/cifs init
It should output a JSON string containing "status": "Success"
. This command is also run by Kubernetes itself when the cifs plugin is detected on the file system.
The plugin takes the CIFS username and password from a Kubernetes Secret. To create the secret, you first have to convert your username and password to base64 encoding:
echo -n username | base64
echo -n password | base64
Then, create a file secret.yml
and use the ouput of the above commands as username and password:
apiVersion: v1
kind: Secret
metadata:
name: cifs-secret
namespace: default
type: fstab/cifs
data:
username: 'ZXhhbXBsZQ=='
password: 'bXktc2VjcmV0LXBhc3N3b3Jk'
Apply the secret:
kubectl apply -f secret.yml
You can check if the secret was installed successfully using kubectl describe secret cifs-secret
.
Next, create a file pod.yml
with a test pod (replace //server/share
with the network path of your CIFS share):
apiVersion: v1
kind: Pod
metadata:
name: busybox
namespace: default
spec:
containers:
- name: busybox
image: busybox
command:
- sleep
- "3600"
imagePullPolicy: IfNotPresent
volumeMounts:
- name: test
mountPath: /data
volumes:
- name: test
flexVolume:
driver: "fstab/cifs"
fsType: "cifs"
secretRef:
name: "cifs-secret"
options:
networkPath: "//server/share"
mountOptions: "dir_mode=0755,file_mode=0644,noperm"
Start the pod:
kubectl apply -f pod.yml
You can verify that the volume was mounted successfully using kubectl describe pod busybox
.
If everything is fine, start a shell inside the container to see if it worked:
kubectl exec -ti busybox /bin/sh
Inside the container, you should see the CIFS share mounted to /data
.