repository·master·Indexed 23 days ago
https://github.com/kubernetes-csi/csi-driver-nfsThe NFS CSI Driver enables Kubernetes to interface with NFSv3 and NFSv4 servers, supporting both static and dynamic volume provisioning via subdirectories. It provides features such as volume cloning (v4.3.0+), high-availability controller configurations, and fsGroupPolicy support for Kubernetes 1.20+ clusters. The driver can be installed and customized using Helm 3, allowing for multiple storage classes and custom driver identities.
The NFS CSI Driver allows Kubernetes clusters to access existing NFSv3 or NFSv4 servers. It supports dynamic provisioning of Persistent Volumes (PVs) via Persistent Volume Claims (PVCs) by automatically creating new subdirectories on the configured NFS server.
Key Details:
nfs.csi.k8s.ioTo run multiple controller replicas (controller.replicas > 1), you must ensure that replicas do not co-locate on the same node. This is because the controller uses hostNetwork: true and its liveness-probe sidecar binds to a fixed host port (controller.livenessProbe.healthPort, default 29652). If two pods attempt to run on the same node, the second pod's sidecar will fail to bind the port, causing a CrashLoopBackOff.
Requirements for HA:
controller.replicas.podAntiAffinity using the kubernetes.io/hostname topology key to prevent co-location.Note: If you see leader-election log lines like "Failed to update lease optimistically, falling back to slow path" in the active controller, this is normal behavior when multiple candidates are racing for the same lease and does not indicate a bug.
Single-node clusters: For development clusters like k3d or kind, you must keep controller.replicas=1.
controller:
replicas: 2
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- topologyKey: kubernetes.io/hostname
labelSelector:
matchLabels:
app: csi-nfs-controllerThe subDir parameter in StorageClass supports dynamic variable substitution. If the subDir value contains any of the following strings, they will be replaced with the corresponding metadata from the PVC or PV:
${pvc.metadata.name}${pvc.metadata.namespace}${pv.metadata.name}After installation, verify that the controller and node pods are running in the kube-system namespace using the following commands:
kubectl -n kube-system get pod -o wide -l app=csi-nfs-controller
kubectl -n kube-system get pod -o wide -l app=csi-nfs-nodeSuccessful pods should show a Running status with all containers (e.g., 4/4 for controller, 3/3 for node) being READY.
kubectl -n kube-system get pod -o wide -l app=csi-nfs-controller
kubectl -n kube-system get pod -o wide -l app=csi-nfs-nodeTo remove the NFS CSI driver version 3.1.0 from your cluster, execute the uninstallation script via a remote curl command.
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v3.1.0/deploy/uninstall-driver.sh | bash -s v3.1.0 --You can install the NFS CSI driver on a Kubernetes cluster using one of two methods: a remote installation via a shell script or a local installation from a cloned repository.
Use curl to fetch and execute the installation script directly from GitHub.
Clone the repository, navigate to the directory, and run the installation script locally.
After running the installation, verify that the controller and node pods are running by checking their status in the kube-system namespace.
# Remote install
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v4.13.1/deploy/install-driver.sh | bash -s v4.13.1 --
# Local install
git clone https://github.com/kubernetes-csi/csi-driver-nfs.git
cd csi-driver-nfs
./deploy/install-driver.sh v4.13.1 local
# Check pods status
kubectl -n kube-system get pod -o wide -l app=csi-nfs-controller
kubectl -n kube-system get pod -o wide -l app=csi-nfs-nodeTo build the project from source, clone the repository into your $GOPATH/src/sigs.k8s.io/ directory and use make to compile the driver. Before submitting code, run make verify to execute verification tests.
If you modify configuration files within the charts directory, you must update the Helm chart package using the following command:
To use the NFS CSI Driver for dynamic provisioning, create a StorageClass with the nfs.csi.k8s.io provisioner. You must specify the server (NFS server address) and share (NFS export path) in the parameters section.
Note: If you need to provide mountOptions for the DeleteVolume operation, you must also include a provisioner secret using the following keys:
csi.storage.k8s.io/provisioner-secret-namecsi.storage.k8s.io/provisioner-secret-namespaceapiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: nfs-csi
provisioner: nfs.csi.k8s.io
parameters:
server: nfs-server.default.svc.cluster.local
share: /
# csi.storage.k8s.io/provisioner-secret is only needed for providing mountOptions in DeleteVolume
# csi.storage.k8s.io/provisioner-secret-name: "mount-options"
# csi.storage.k8s.io/provisioner-secret-namespace: "default"
reclaimPolicy: Delete
volumeBindingMode: Immediate
allowVolumeExpansion: true
mountOptions:
- nfsvers=4.1You can install the NFS CSI driver on a Kubernetes cluster using kubectl via two methods: a remote installation using a shell script from GitHub, or a local installation using a cloned repository.
Use curl to fetch and execute the installation script directly from the repository.
Clone the repository, navigate to the directory, and run the installation script locally.
After installation, verify that the controller and node pods are running in the kube-system namespace using the following commands:
# Remote install
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v4.12.0/deploy/install-driver.sh | bash -s v4.12.0 --
# Local install
git clone https://github.com/kubernetes-csi/csi-driver-nfs.git
cd csi-driver-nfs
./deploy/install-driver.sh v4.12.0 local
# Check pods status
kubectl -n kube-system get pod -o wide -l app=csi-nfs-controller
kubectl -n kube-system get pod -o wide -l app=csi-nfs-nodeTo remove the NFS CSI driver version 4.2.0 from your cluster, use either the remote uninstall script or the local uninstall script.
Execute the uninstall script directly from GitHub using curl.
Clone the repository, check out the specific version v4.2.0, and run the local uninstall script.
# Remote uninstall
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v4.2.0/deploy/uninstall-driver.sh | bash -s v4.2.0 --
# Local uninstall
git clone https://github.com/kubernetes-csi/csi-driver-nfs.git
cd csi-driver-nfs
git checkout v4.2.0
./deploy/uninstall-driver.sh v4.2.0 localTo remove version 3.0.0 of the NFS CSI driver, execute the uninstallation script via curl.
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v3.0.0/deploy/uninstall-driver.sh | bash -s v3.0.0 --You can install the NFS CSI driver v4.13.4 on your Kubernetes cluster using two methods: a remote installation via a shell script or a local installation from the repository source.
Use curl to fetch and execute the installation script directly from GitHub.
Clone the repository, navigate to the directory, and run the installation script locally.
After installation, verify the status of the controller and node pods to ensure they are running.
# Option#1. remote install
curl -skSL https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/v4.13.4/deploy/install-driver.sh | bash -s v4.13.4 --
# Option#2. local install
git clone https://github.com/kubernetes-csi/csi-driver-nfs.git
cd csi-driver-nfs
./deploy/install-driver.sh v4.13.4 local
# check pods status
kubectl -n kube-system get pod -o wide -l app=csi-nfs-controller
kubectl -n kube-system get pod -o wide -l app=csi-nfs-node