All the below steps are expected to be performed by an organization administrator.
It is a common practice to create a temporary bootstrap cluster which is then used to provision a management cluster on the Cloud Director (infrastructure provider).
Choose one of the options below to set up a management cluster on VMware Cloud Director:
- CSE provisioned TKG cluster as a bootstrap cluster to further create a Management cluster in VCD tenant organization.
- Kind as a bootstrap cluster to create Management cluster in VCD tenant organization.
The next step is to initialize the bootstrap cluster with the Cluster API
- Set up the clusterctl on the cluster
- Initialize the Cluster API and CAPVCD on the cluster
- Apply CRS definitions to ensure CPI, CNI, and CSI are installed on the (children) workload clusters
- Wait until
kubectl get pods -A
shows below pods in Running state-
kubectl get pods -A NAMESPACE NAME READY STATUS capi-kubeadm-bootstrap-system capi-kubeadm-bootstrap-controller-manager-7dc44947-v5nlv 1/1 Running capi-kubeadm-control-plane-system capi-kubeadm-control-plane-controller-manager-cb9d954f5-ct5cp 1/1 Running capi-system capi-controller-manager-7594c7bc57-smjtg 1/1 Running capvcd-system capvcd-controller-manager-769d64d4bf-54bf4 1/1 Running
-
Now that bootstrap cluster is ready, you can use Cluster API to create a multi control-plane workload cluster
- Generate the cluster manifest and apply it on the bootstrap cluster to create a brand-new workload cluster.
- Apply CRS labels on the workload cluster.
- The next step is to enable add-ons (CPI, CSI) on the workload cluster to access VCD resources
- Transform this workload cluster into management cluster by initializing it with Cluster API and CAPVCD.
- This cluster is now a fully functional multi-control plane management cluster; you can use this to create and manage multiple workload clusters.
These workflows need to be run from the bootstrap cluster (the parent of the management cluster).
In the kubectl
commands specified in the below workflows, update the namespace
parameter to the value default
and kubeconfig
to the value of bootstrap cluster's admin Kubeconfig
This is an advanced workflow to enable Cloud Director tenant users to deploy workload clusters from a single management cluster in an isolated manner.
Refer to enable multitenancy on the management cluster for more details
Configuring Machine Health Checks on the management cluster will instruct Cluster API to detect unhealthy machines of a given cluster and remediate them.
Refer to Machine Health Checks for more details.