Developer Mode Installation

Requirements​

• The Kubernetes node must pass the host-check
• If the Kubelet's RootDir is not /var/lib/kubelet, you must create a bind mount to /var/lib/kubelet as follows:

  KUBELET_ROOT_DIR="path to your kubelet root dir"
  echo "${KUBELET_ROOT_DIR} /var/lib/kubelet none bind 0 0" >> /etc/fstab
  mkdir -p /var/lib/kubelet && mount -a
  
• Multus is installed across your cluster and a corresponding NetworkAttachmentDefinition CRD is created.
• The Harvester Chart already contains the Kubevirt and Longhorn

Installation​

For development purpose, Harvester can be installed on a Kubernetes cluster by using the Helm CLI.

Please refer to the Harvester Helm chart for more details on installing and configuring the Helm chart.

1. Create the cattle-system namespace

   kubectl create ns cattle-system
   

2. Add the rancher-latest helm repo

   helm repo add rancher-latest https://releases.rancher.com/server-charts/latest
   

3. Install Rancher chart

   helm install rancher rancher-latest/rancher \
   --namespace cattle-system \
   --set tls=external \
   --set rancherImagePullPolicy=IfNotPresent \
   --set rancherImage=rancher/rancher \
   --set rancherImageTag=v2.6.3-harvester1 \
   --set noDefaultAdmin=false \
   --set features="multi-cluster-management=false\,multi-cluster-management-agent=false" \
   --set useBundledSystemChart=true \
   --set bootstrapPassword=admin
   

4. Change the 'status.provider' of the local cluster object to "harvester"

   kubectl edit clusters.management.cattle.io local
   

5. Clone the GitHub repository

   git clone https://github.com/harvester/harvester.git --depth=1
   

6. Go to the Helm chart

   cd harvester/deploy/charts
   

7. Create the harvester-system namespace

   kubectl create ns harvester-system
   

8. Install the Harvester crd chart

   helm install harvester-crd ./harvester-crd --namespace harvester-system
   

9. Install the Harvester chart

   ## In order to use the service type LoadBalancer and create a vip in control-plane nodes, we need to enable kubevip.
   VIP_IP="replace with your vip ip, such as 192.168.5.10"
   VIP_NIC="replace with your vip interface name, such as eth0"
   helm install harvester ./harvester --namespace harvester-system \
   --set harvester-node-disk-manager.enabled=true \
   --set harvester-network-controller.enabled=true \
   --set harvester-load-balancer.enabled=true \
   --set kube-vip.enabled=true \
   --set kube-vip.config.vip_interface=${VIP_NIC} \
   --set kube-vip.config.vip_address=${VIP_IP} \
   --set service.vip.enabled=true \
   --set service.vip.ip=${VIP_IP}
   

   ## In some Kubernetes distributions (such as kubeadm), we need to modify the kube-vip nodeSelector to match the control-plane nodes.
   --set kube-vip.nodeSelector."node-role\.kubernetes\.io/master"=""
   

10. Expose Harvester UI

    ## Refer to https://kube-vip.chipzoller.dev/docs/usage/cloud-provider/. Add `cidr-cattle-system: ${VIP_IP}/32` to kubevip configMap.
    kubectl -n kube-system edit cm kubevip
    
    ## Change the rancher service type from ClusterIP to LoadBalancer, and then you can access Harvester UI via https://${VIP_IP}.
    kubectl -n cattle-system edit svc rancher
    

DigitalOcean Test Environment​

You can create a test Kubernetes environment in Rancher using DigitalOcean as a cloud provider, which supports nested virtualization.

We recommend using a 8 core, 16 GB RAM droplet, which will have nested virtualization enabled by default.

This screenshot shows how to create a Rancher node template that would allow Rancher to provision such a node in DigitalOcean:

For more information on how to launch DigitalOcean nodes with Rancher, refer to the Rancher documentation.