Distributed block storage for Kubernetes
Switch branches/tags
Nothing to show
Clone or download
yasker Update README.md
Add RKE Flexvolume plugin location.
Latest commit 4bceb56 Jun 20, 2018
Failed to load latest commit information.
deploy Update maager to rancher/longhorn-manager:010fe60 Apr 14, 2018
examples examples: Add deployment example Mar 26, 2018
LICENSE Initial commit Apr 14, 2017
README.md Update README.md Jun 20, 2018



Longhorn is a distributed block storage system for Kubernetes. Longhorn is lightweight, reliable, and easy-to-use. You can deploy Longhorn on an existing Kubernetes cluster with one simple command. Once Longhorn is deployed, it adds persistent volume support to the Kubernetes cluster.

Longhorn implements distributed block storage using containers and microservices. Longhorn creates a dedicated storage controller for each block device volume and sychronously replicates the volume across multiple replicas stored on multiple nodes. The storage controller and replicas are themselves orchestrated using Kubernetes. Longhorn supports snapshots, backups, and even allows you to schedule recurring snapshots and backups!

You can read more details of Longhorn and its design here.

Longhorn is a work in progress. We appreciate your comments as we continue to work on it!

Source Code

Longhorn is 100% open source software. Project source code is spread across a number of repos:

  1. Longhorn Engine -- Core controller/replica logic https://github.com/rancher/longhorn-engine
  2. Longhorn Manager -- Longhorn orchestration, includes Flexvolume driver for Kubernetes https://github.com/rancher/longhorn-manager
  3. Longhorn UI -- Dashboard https://github.com/rancher/longhorn-ui


Longhorn v0.2 Demo

Deploy on Kubernetes


  1. Docker v1.13+
  2. Kubernetes v1.8+
  3. Make sure curl, findmnt, grep, awk and blkid has been installed in all nodes of the Kubernetes cluster.
  4. Make sure open-iscsi has been installed in all nodes of the Kubernetes cluster. For GKE, recommended Ubuntu as guest OS image since it contains open-iscsi already.


Create the deployment of Longhorn in your Kubernetes cluster is easy. For most Kubernetes setup (except GKE), you will only need to run the following command to install Longhorn:

kubectl create -f https://raw.githubusercontent.com/rancher/longhorn/master/deploy/longhorn.yaml

For Google Kubernetes Engine (GKE) users, see here before proceed.

Longhorn Manager and Longhorn Driver will be deployed as daemonsets in a separate namespace called longhorn-system, as you can see in the yaml file.

When you see those pods has started correctly as follows, you've deployed the Longhorn successfully.

# kubectl -n longhorn-system get pod
NAME                                                  READY     STATUS    RESTARTS   AGE
longhorn-flexvolume-driver-4dnx6                      1/1       Running   0          1d
longhorn-flexvolume-driver-cqwj5                      1/1       Running   0          1d
longhorn-flexvolume-driver-deployer-bc7b95b5b-sb9kr   1/1       Running   0          1d
longhorn-flexvolume-driver-q9h4f                      1/1       Running   0          1d
longhorn-manager-dkdn9                                1/1       Running   0          2h
longhorn-manager-l6npd                                1/1       Running   0          2h
longhorn-manager-v4fz8                                1/1       Running   0          2h
longhorn-ui-58796c68d-db4t6                           1/1       Running   0          1h

Access the UI

Use kubectl -n longhorn-system get svc to get the external service IP for UI:

NAME                TYPE           CLUSTER-IP      EXTERNAL-IP      PORT(S)        AGE
longhorn-backend    ClusterIP   <none>           9500/TCP       58m
longhorn-frontend   LoadBalancer   80:30697/TCP   58m

If the Kubernetes Cluster supports creating LoadBalancer, user can then use EXTERNAL-IP( in the case above) of longhorn-frontend to access the Longhorn UI. Otherwise the user can use <node_ip>:<port> (port is 30697 in the case above) to access the UI.

Longhorn UI would connect to the Longhorn Manager API, provides the overview of the system, the volume operations, and the snapshot/backup operations. It's highly recommended for the user to check out Longhorn UI.

Notice the current UI is unauthenticated.

How to use the Longhorn Volume in your pod

There are serveral ways to use the Longhorn volume.

Pod with Longhorn volume

The following YAML file shows the definition of a pod that makes the Longhorn attach a volume to be used by the pod.

apiVersion: v1
kind: Pod
  name: volume-test
  namespace: default
  - name: volume-test
    image: nginx:stable-alpine
    imagePullPolicy: IfNotPresent
    - name: voll
      mountPath: /data
    - containerPort: 80
  - name: voll
      driver: "rancher.io/longhorn"
      fsType: "ext4"
        size: "2Gi"
        numberOfReplicas: "3"
        staleReplicaTimeout: "20"
        fromBackup: ""

Notice this field in the YAML file: flexVolume.driver "rancher.io/longhorn". It specifies that the Longhorn FlexVolume plug-in should be used. There are some option fields in options the user can fill in.

Option Required Description
size Yes Specify the capacity of the volume in longhorn and the unit should be G
numberOfReplicas Yes The number of replicas (HA feature) for volume in this Longhorn volume
fromBackup No Optional. Must be a Longhorn Backup URL. Specify where the user want to restore the volume from.

Storage class

Longhorn supports dynamic provisioner function, which can create PV automatically for the user according to the spec of storage class and PVC. The user needs to create a new storage class in order to use it. The storage class example can be downloaded from here

kind: StorageClass
apiVersion: storage.k8s.io/v1
  name: longhorn
provisioner: rancher.io/longhorn
  numberOfReplicas: "3"
  staleReplicaTimeout: "30"
  fromBackup: ""

Then user can create a PVC directly. For example:

apiVersion: v1
kind: PersistentVolumeClaim
  name: longhorn-volv-pvc
    - ReadWriteOnce
  storageClassName: longhorn
      storage: 2Gi

Then use it in the pod:

apiVersion: v1
kind: Pod
  name: volume-test
  namespace: default
  - name: volume-test
    image: nginx:stable-alpine
    imagePullPolicy: IfNotPresent
    - name: volv
      mountPath: /data
    - containerPort: 80
  - name: volv
      claimName: longhorn-volv-pvc

Setup a TESTING ONLY NFS server for storing backups

Longhorn supports backing up mechanisms to export the user data out of the Longhorn system. Currently Longhorn supports backing up to a NFS server. In order to use this feature, you need to have a NFS server running and accessible in the Kubernetes cluster. Here we provide a simple way to setup a testing NFS server.

WARNING: This NFS server won't save any data after you delete it. It's for TESTING ONLY.

kubectl create -f deploy/example-backupstore.yaml

It will create a simple NFS server in the default namespace, which can be addressed as longhorn-test-nfs-svc.default for other pods in the cluster.

After this script completes, using the following URL as the Backup Target in the Longhorn setting:


Open Longhorn UI, go to Setting, fill the Backup Target field with the URL above, click Save. Now you should able to use the backup feature of Longhorn.

Google Kubernetes Engine

The configuration yaml will be slight different for Google Kubernetes Engine (GKE):

  1. GKE requires user to manually claim himself as cluster admin to enable RBAC. User need to execute following command before create the Longhorn system using yaml files.
kubectl create clusterrolebinding cluster-admin-binding --clusterrole=cluster-admin --user=<name@example.com>

In which name@example.com is the user's account name in GCE, and it's case sensitive. See here for details.

  1. The default Flexvolume plugin directory is different with GKE 1.8+, which is at /home/kubernetes/flexvolume. User need to use following command instead:
kubectl create -f https://raw.githubusercontent.com/rancher/longhorn/master/deploy/longhorn-gke.yaml

User can also customerize the Flexvolume directory in the last part of the Longhorn system deployment yaml file, e.g.:

          - name: FLEXVOLUME_DIR
            value: "/home/kubernetes/flexvolume/"

See Troubleshooting for details.

Uninstall Longhorn

In order to uninstall Longhorn, user need to remove all the volumes first:

kubectl -n longhorn-system delete lhv --all

After confirming all the volumes are removed, then Longhorn can be easily uninstalled using:

kubectl delete -f https://raw.githubusercontent.com/rancher/longhorn/master/deploy/longhorn.yaml


Volume can be attached/detached from UI, but Kubernetes Pod/StatefulSet etc cannot use it

Check if volume plugin directory has been set correctly.

By default, Kubernetes use /usr/libexec/kubernetes/kubelet-plugins/volume/exec/ as the directory for volume plugin drivers, as stated in the official document.

But some vendors may choose to change the directory due to various reasons. For example, GKE uses /home/kubernetes/flexvolume, and RKE uses /var/lib/kubelet/volumeplugins.

User can find the correct directory by running ps aux|grep kubelet on the host and check the --volume-plugin-dir parameter. If there is none, the default /usr/libexec/kubernetes/kubelet-plugins/volume/exec/ will be used.


Copyright (c) 2014-2018 Rancher Labs, Inc.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at


Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.