Users

This guide will give you a quick start on installing and setting up the ClusterLink on a Kubernetes cluster.

Prerequisites

Before you start, you must have access to a Kubernetes cluster. For example, you can set up a local environment using the kind project.

Installation

1. To install ClusterLink on Linux or Mac, use the installation script:

   curl -L https://github.com/clusterlink-net/clusterlink/releases/latest/download/clusterlink.sh | sh -
   

2. Check the installation by running the command:

   clusterlink --version
   

Setup

To set up ClusterLink on a Kubernetes cluster, follow these steps:

1. Create the fabric’s CA certificate and private key:

   clusterlink create fabric --name <fabric_name>
   

   The ClusterLink fabric is defined as all K8s clusters (peers) that install ClusterLink gateways and can share services between the clusters, enabling communication among those services. This command will create the CA files cert.pem and key.pem in a directory named <fabric_name>. The --name option is optional, and by default, “default_fabric” will be used.

2. Create a peer (cluster) certificate:

   clusterlink create peer-cert --name <peer_name> --fabric <fabric_name>
   

   This command will create the certificate files cert.pem and key.pem in a directory named <fabric_name>/<peer_name>. The --path <path> flag can be used to change the directory location. The --name option is optional, and by default, “default_fabric” will be used.

3. Install ClusterLink deployment operator:

   clusterlink peer deploy --autostart --name <peer_name> --fabric <fabric_name>
   

   This command will deploy the ClusterLink operator on the clusterlink-operator namespace and convert the peer certificates to secrets in this namespace. The command assumes that kubectl is set to the correct peer (K8s cluster) and that the certificates were created by running the previous command on the same working directory. If they were not, use the flag --path <path> for pointing to the working directory that was used in the previous command. The --fabric option is optional, and by default, “default_fabric” will be used. The --autostart option will deploy the ClusterLink components in the clusterlink-system namespace, and the ClusterLink project will start running in the cluster.

To deploy ClusterLink on another cluster, please repeat steps 2-3 in a console with access to the cluster.

Additional configurations

• Setting ClusterLink namespace:

  clusterlink peer deploy --autostart --name <peer_name> --fabric <fabric_name> --namespace <namespace>
  

  The --namespace parameter determines the namespace where the ClusterLink components are deployed. Note that you must set --autostart, and the namespace should already exist.

• Setting ClusterLink ingress type:

  clusterlink peer deploy --autostart --name <peer_name> --fabric <fabric_name> --ingress <ingress_type>
  

  The --ingress parameter controls the ClusterLink ingress type, with LoadBalancer being the default. If you’re using a kind cluster, replace <ingress_type> with NodePort. For a cluster running in a cloud environment, use LoadBalancer. Note that you must set --autostart when you use this parameter.

• Full configuration setting using a ClusterLink K8s custom resource object:

  kubectl apply -f - <<EOF
  apiVersion: clusterlink.net/v1alpha1
  kind: ClusterLink
  metadata:
      namespace: clusterlink-operator
      name: <peer_name>
  spec:
      ingress:
          type: <ingress_type>
      namespace: clusterlink-system
  EOF
  

  After the operator is installed, you can deploy ClusterLink by applying a ClusterLink object as shown above. The Clusterlink operator will create the ClusterLink gateway components in the clusterlink-system namespace. For more details and information about the ClusterLink custom resource, refer to the operator documentation.

Try it out

Check out the ClusterLink Tutorials for setting up multi-cluster connectivity for applications using two or more clusters.