KaaS – Cluster API installation

Prerequisites

1. Tools

To install cluster API on a Kubernetes cluster it is required to have these tools installed on the machine used to administer :

1. kubectl
2. clusterctl
3. git
4. kctx (optional)

If one of these tool is missing, please follow this guide here

2. Cluster connection

Test that you are correctly connected to the cluster you want to install Cluster API on.

If you are not connected to the good cluster, switch your kubectl to the good kubeconfig file.

For that you can use kctx or any other preferred way. How to install kctx

# kubectl get nodes

NAME        STATUS   ROLES                  AGE   VERSION
mstr-atom   Ready    control-plane,master   61m   v1.22.9+vmware.1
node-3v9t   Ready    <none>                 58m   v1.22.9+vmware.1

3. VCD Account

A VCD Account will be required for Cluster API, this account will be used for all operations made by cluster API (cluster creation, storage provisionning, load balancer operations).

You can use default roles or create a specific one with these rights :

• User > Manage user’s own API token
• vApp > Preserve ExtraConfig Elements during OVA Import and Export (follow the KB to enable this right on VCD)
• Catalog > View Shared Catalogs from Other Organizations
• Gateway > View Gateway
• Gateway Services > NAT Configure, LoadBalancer Configure
• Rights from the default vApp Author role
• Right ‘Full Control: VMWARE:CAPVCDCLUSTER’
• Organization VDC > Create a Shared Disk

Installation

Prepare the Cluster API Provider for vCloud Director

Cluster API a besoin d’un provider afin de pouvoir communiquer avec l’API de vCloud Director afin d’orchestrer le déploiement et de gérer le cycle de vie des clusters.

Ce provider est fourni par VMware et est disponible sur un reposository GIT que l’on va utiliser pour l’installation de Cluster API.

Le principe est le suivant :

1. Clonez le repository GIT
2. Récupérez les fichiers de template et de configuration
3. Déclarez le provider dans la configuration de cluster api

git clone --branch v1.1.0 https://github.com/vmware/cluster-api-provider-cloud-director.git

Prepare templates and configuration files.

mkdir -p infrastructure-vcd/v1.1.0/
cp -r cluster-api-provider-cloud-director/templates/* ./infrastructure-vcd/v1.1.0/
cp cluster-api-provider-cloud-director/metadata.yaml ./infrastructure-vcd/v1.1.0/
mkdir ~/.cluster-api/
cp infrastructure-vcd/v1.1.0/clusterctl.yaml ~/.cluster-api/clusterctl.yaml

Declare the provider in the Cluster API configuration

1. Edit the ~/.cluster-api/clusterctl.yaml file.
2. Modify the url parameter for the vCloud Director (vcd) provider.

Here’s a general example of what the relevant section of your clusterctl.yaml file might look like :

providers:
  - name: "vcd"
    url: "/root/infrastructure-vcd/v1.1.0/infrastructure-components.yaml"
    type: "InfrastructureProvider"

Make sure to replace "/path/to/infrastructure-components.yaml" with the actual file path to your infrastructure-components.yaml file. This configuration tells Cluster API where to find the provider and its configuration.

Cluster API Installation

1. Launch clusterctl init command

You should received this message :

Your management cluster has been initialized successfully!

clusterctl init --core cluster-api:v1.4.0 -b kubeadm:v1.4.0 -c kubeadm:v1.4.0 -i vcd:v1.1.0

Apply the CRS (ClusterResourceSet)

The ClusterResourceSet is an object from the Cluster API, and it allows for the automatic creation of resources on the child clusters that will be deployed. In our case, we are using it to deploy automatically:

• CNI Antrea
• CSI for vCloud Director (for Persistent Storage)
• CPI for vCloud Director (for LB)

If you copied the files to a different directory in the previous steps, you can modify the first line to provide the correct location.

CRS="infrastructure-vcd/v1.1.0/crs/"
kubectl create configmap antrea-crs-cm --from-file=${CRS}/cni/antrea.yaml
kubectl apply -f ${CRS}/cni/antrea-crs.yaml

kubectl create configmap cloud-director-crs-cm --from-file=${CRS}/cpi/cloud-director-ccm.yaml
kubectl apply -f ${CRS}/cpi/cloud-director-crs.yaml

kubectl create configmap csi-controller-crs-cm --from-file=${CRS}/csi/csi-controller-crs.yaml
kubectl create configmap csi-node-crs-cm --from-file=${CRS}/csi/csi-node-crs.yaml
kubectl create configmap csi-driver-crs-cm --from-file=${CRS}/csi/csi-driver.yaml
kubectl apply -f ${CRS}/csi/csi-crs.yaml

All PODS and configurations will be automatically applied to the cluster.

Please wait for all these PODS to be in the “Running” status.

If an error occurs, please refer to the page. KaaS Standard – Troubleshooting

# kubectl get pods -A

NAMESPACE                           NAME                                                            READY   STATUS    RESTARTS   AGE
capi-kubeadm-bootstrap-system       capi-kubeadm-bootstrap-controller-manager-7dc44947-hrmvc        1/1     Running   0          36m
capi-kubeadm-control-plane-system   capi-kubeadm-control-plane-controller-manager-cb9d954f5-r8w54   1/1     Running   0          36m
capi-system                         capi-controller-manager-7594c7bc57-jr75r                        1/1     Running   0          37m
capvcd-system                       capvcd-controller-manager-89758d745-kw4sm                       1/1     Running   0          13s

Next steps

Next you have to deploy your cluster, it can be a management cluster or a workload cluster depending of your need :

• Deploy a management cluster : KaaS Standard – Management Cluster
• Deploy a workload cluster : KaaS Advanced – Deploy workload cluster