Installation on a single VM

Nephio Installation in a sandbox VM

This page is draft and the separation of the content to different categories is not clearly done.

In this guide, you will set up Nephio running in a single VM with:

  • Management Cluster: kind
  • Cluster Provisioner: Cluster API
  • Workload Clusters: kind
  • Gitops Tool: ConfigSync
  • Git Provider: Gitea running in the Nephio management cluster will be the git provider for cluster deployment repositories. Some external repositories will be on GitHub.
  • Web UI Auth: None
  • Ingress/Load Balancer: MetalLB, but only internally to the VM.

Provisioning Your Sandbox VM

In addition to the general prerequisites, you will need:

  • Access to a Virtual Machine provided by an hypervisor ( VirtualBox, Libvirt) and running an OS supported by Nephio (Ubuntu 20.04/22.04, Fedora 34) with a minimum of 16 vCPUs and 32 GB in RAM.
  • Kubernetes IN Docker (kind) installed and set up your workstation.

Provisioning Your Management Cluster

The Cluster API services require communication with the Docker socket for creation of workload clusters. The command below creates an All-in-One Nephio management cluster through the KinD tool, mapping the /var/run/docker.sock socket file for Cluster API communication.

cat << EOF | kind create cluster --config=-
kind: Cluster
  - role: control-plane
    image: kindest/node:v1.27.1
      - hostPath: /var/run/docker.sock
        containerPath: /var/run/docker.sock

Gitea Installation

While you may use other Git providers as well, Gitea is required in the R2 setup. To install Gitea, use kpt. From your nephio-install directory, run:

kpt pkg get --for-deployment
kpt fn render gitea
kpt live init gitea
kpt live apply gitea --reconcile-timeout 15m --output=table

Common Dependencies

There are a few dependencies that are common across most installations, and do not require any installation-specific setup. You should install these next, as described in the common dependencies documentation.

Common Components

With the necessary dependencies now installed, you can now install the essential Nephio components. This is documented in the common components documentation.

Provisioning Cluster API

For managing the Kubernetes cluster infrastructure, it is necessary to install Cluster API project. This package depends on cert-manager project to generate certificates.

kpt pkg get --for-deployment
kpt fn render cert-manager
kpt live init cert-manager
kpt live apply cert-manager --reconcile-timeout 15m --output=table

Once cert-manager is installed, you can proceed with the installation of Cluster API components

kpt pkg get --for-deployment
kpt fn render cluster-capi
kpt live init cluster-capi
kpt live apply cluster-capi --reconcile-timeout 15m --output=table

Cluster API uses infrastructure providers to provision cloud resources required by the clusters. You can manage local resources with the Docker provider, it can be installed with the followed package.

kpt pkg get --for-deployment
kpt fn render cluster-capi-infrastructure-docker
kpt live init cluster-capi-infrastructure-docker
kpt live apply cluster-capi-infrastructure-docker --reconcile-timeout 15m --output=table

The last step is required for defining cluster, machine and kubeadmin templates for controller and worker docker machines. These templates define the kubelet args, etcd and coreDNS configuration and image repository as other things.

kpt pkg get --for-deployment
kpt fn render cluster-capi-kind-docker-templates
kpt live init cluster-capi-kind-docker-templates
kpt live apply cluster-capi-kind-docker-templates --reconcile-timeout 15m --output=table