README
¶
JuiceFS CSI Driver
The JuiceFS Container Storage Interface (CSI) Driver implements the CSI specification for container orchestrators to manage the lifecycle of JuiceFS file system.
Prerequisites
- Kubernetes 1.14+
Installation
Installation with Helm
Prerequisites
- Helm 3.1.0+
Install Helm
Helm is a tool for managing Kubernetes charts. Charts are packages of pre-configured Kubernetes resources.
To install Helm, refer to the Helm install guide and ensure that the helm binary is in the PATH of your shell.
Using Helm To Deploy
- Prepare a
values.yamlfile with access information about metadata engine (e.g. Redis) and object storage. If already formatted a volume, onlynameandmetaurlis required. You can specify CPU/memory limits and requests of mount pod for pods using this driver.
storageClasses:
- name: juicefs-sc
enabled: true
reclaimPolicy: Retain
backend:
name: "<name>"
metaurl: "<meta-url>"
storage: "<storage-type>"
accessKey: "<access-key>"
secretKey: "<secret-key>"
bucket: "<bucket>"
mountPod:
resources:
limits:
cpu: "<cpu-limit>"
memory: "<memory-limit>"
requests:
cpu: "<cpu-request>"
memory: "<memory-request>"
- Install
helm repo add juicefs-csi-driver https://juicedata.github.io/juicefs-csi-driver/
helm repo update
helm upgrade juicefs-csi-driver juicefs-csi-driver/juicefs-csi-driver --install -f ./values.yaml
- Check the deployment
- Check pods are running: the deployment will launch a
StatefulSetnamedjuicefs-csi-controllerwith replica1and aDaemonSetnamedjuicefs-csi-node, so runkubectl -n kube-system get pods -l app.kubernetes.io/name=juicefs-csi-drivershould seen+1(wherenis the number of worker nodes of the Kubernetes cluster) pods is running. For example:
$ kubectl -n kube-system get pods -l app.kubernetes.io/name=juicefs-csi-driver
NAME READY STATUS RESTARTS AGE
juicefs-csi-controller-0 3/3 Running 0 22m
juicefs-csi-node-v9tzb 3/3 Running 0 14m
- Check secret:
kubectl -n kube-system describe secret juicefs-sc-secretwill show the secret with abovebackendfields invalues.yaml:
Name: juicefs-sc-secret
Namespace: kube-system
Labels: app.kubernetes.io/instance=juicefs-csi-driver
app.kubernetes.io/managed-by=Helm
app.kubernetes.io/name=juicefs-csi-driver
app.kubernetes.io/version=0.7.0
helm.sh/chart=juicefs-csi-driver-0.1.0
Annotations: meta.helm.sh/release-name: juicefs-csi-driver
meta.helm.sh/release-namespace: default
Type: Opaque
Data
====
access-key: 0 bytes
bucket: 47 bytes
metaurl: 54 bytes
name: 4 bytes
secret-key: 0 bytes
storage: 2 bytes
- Check storage class:
kubectl get sc juicefs-scwill show the storage class like this:
NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE
juicefs-sc csi.juicefs.com Retain Immediate false 69m
Installation with kubectl
Deploy the driver:
kubectl apply -f https://raw.githubusercontent.com/juicedata/juicefs-csi-driver/master/deploy/k8s.yaml
If the CSI driver couldn't be discovered by Kubernetes and the error like this: driver name csi.juicefs.com not found in the list of registered CSI drivers, check the root directory path of kubelet. Run the following command on any non-master node in your Kubernetes cluster:
ps -ef | grep kubelet | grep root-dir
If the result isn't empty, modify the CSI driver deployment k8s.yaml file with the new path and redeploy the CSI driver again.
curl -sSL https://raw.githubusercontent.com/juicedata/juicefs-csi-driver/master/deploy/k8s.yaml | sed 's@/var/lib/kubelet@{{KUBELET_DIR}}@g' | kubectl apply -f -
Replace {{KUBELET_DIR}} with your own --root-dir value in above command.
Upgrade CSI Driver
v0.10+
Juicefs CSI Driver separated JuiceFS client from CSI Driver since v0.10.0, CSI Driver upgrade will not interrupt existing PVs. If CSI Driver version >= v0.10.0, do operations below:
- If you're using
latesttag, simple runkubectl rollout restart -f k8s.yamland make surejuicefs-csi-controllerandjuicefs-csi-nodepods are restarted. - If you have pinned to a specific version, modify your
k8s.yamlto a newer version, then runkubectl apply -f k8s.yaml. - Alternatively, if JuiceFS CSI driver is installed using Helm, you can also use Helm to upgrade it.
If you want to upgrade CSI Driver from v0.9.0 to v0.10.0+, follow "How to upgrade CSI Driver from v0.9.0 to v0.10.0+".
Before v0.10
Upgrade of CSI Driver requires restart the DaemonSet, which has all the JuiceFS client running inside. The restart will cause all PVs become unavailable, so we need to stop all the application pod first.
- Stop all pods using this driver.
- Upgrade driver:
- If you're using
latesttag, simple runkubectl rollout restart -f k8s.yamland make surejuicefs-csi-controllerandjuicefs-csi-nodepods are restarted. - If you have pinned to a specific version, modify your
k8s.yamlto a newer version, then runkubectl apply -f k8s.yaml. - Alternatively, if JuiceFS CSI driver is installed using Helm, you can also use Helm to upgrade it.
- If you're using
- Start all the application pods.
We are working on launching the JuiceFS client in separate container, so the upgrade will NOT interrupt existing PVs. Before that, we can upgrade the JuiceFS client without upgrading the CSI driver, please follow this workaround.
Visit Docker Hub for more versions.
Examples
Before the example, you need to:
- Get yourself familiar with how to setup Kubernetes and how to use JuiceFS file system.
- Make sure JuiceFS is accessible from Kuberenetes cluster. It is recommended to create the file system inside the same region as Kubernetes cluster.
- Install JuiceFS CSI driver following the Installation steps.
Example links
Notes:
- Since JuiceFS is an elastic file system it doesn't really enforce any file system capacity. The actual storage capacity value in PersistentVolume and PersistentVolumeClaim is not used when creating the file system. However, since the storage capacity is a required field by Kubernetes, you must specify the value and you can use any valid value e.g.
10Pifor the capacity. - kustomize 3.x is required to build some examples.
CSI Specification Compatibility
| JuiceFS CSI Driver \ CSI Version | v0.3 | v1.0 |
|---|---|---|
| master branch | no | yes |
Interfaces
The following CSI interfaces are implemented:
- Node Service: NodePublishVolume, NodeUnpublishVolume, NodeGetCapabilities, NodeGetInfo, NodeGetId
- Identity Service: GetPluginInfo, GetPluginCapabilities, Probe
Troubleshooting
If you encounter any issue, please refer to Troubleshooting document.
JuiceFS CSI Driver on Kubernetes
The following sections are Kubernetes specific. If you are a Kubernetes user, use this for driver features, installation steps and examples.
Kubernetes Version Compatibility
JuiceFS CSI Driver is compatible with Kubernetes v1.14+
Container Images
| JuiceFS CSI Driver Version | Image |
|---|---|
| master branch | juicedata/juicefs-csi-driver:latest |
Features
- Static provisioning - JuiceFS file system needs to be created manually first, then it could be mounted inside container as a PersistentVolume (PV) using the driver.
- Mount options - CSI volume attributes can be specified in the PersistentVolume (PV) to define how the volume should be mounted.
- Read write many - Support
ReadWriteManyaccess mode - Sub path - provision PersistentVolume with subpath in JuiceFS file system
- Mount resources - CSI volume attributes can be specified in the PersistentVolume (PV) to define CPU/memory limits/requests of mount pod.
- Config files & env in mount pod - Support set config files and envs in mount pod.
- Dynamic provisioning - allows storage volumes to be created on-demand
Validation
JuiceFS CSI driver has been validated in the following Kubernetes version
| Kubernetes | master |
|---|---|
| v1.19.2 / minikube v1.16.0 | Yes |
| v1.20.2 / minikube v1.16.0 | Yes |
Known issues
The mount option --cache-dir in JuiceFS CSI driver (>=v0.10.0) does not support wildcards currently.
Miscellaneous
Develop
See DEVELOP document.
License
This library is licensed under the Apache 2.0 License.