k0rdent Cluster Manager is part of k0rdent which is focused on delivering a open source approach to providing an enterprise grade multi-cluster kubernetes management solution based entirely on standard open source tooling that works across private or public clouds.

We like to say that Project 0x2A (42) is the answer to life, the universe, and everything ... Or, at least, the Kubernetes sprawl we find ourselves faced with in real life!

Detailed documentation is available in k0rdent Docs

helm install kcm oci://ghcr.io/k0rdent/kcm/charts/kcm --version 0.1.0 -n kcm-system --create-namespace

Then follow the Deploy a cluster deployment guide to create a cluster deployment.

See Install KCM for development purposes.

KCM requires the following:

1. Existing management cluster (minimum required kubernetes version 1.28.0).
2. kubectl CLI installed locally.

Optionally, the following CLIs may be helpful:

1. helm (required only when installing KCM using helm).
2. clusterctl (to handle the lifecycle of the cluster deployments).

Full details on the provider configuration can be found in the k0rdent Docs, see Documentation

export KUBECONFIG=<path-to-management-kubeconfig>

helm install kcm oci://ghcr.io/k0rdent/kcm/charts/kcm --version <kcm-version> -n kcm-system --create-namespace

By default, kcm is being deployed with the following configuration:

apiVersion: k0rdent.mirantis.com/v1alpha1
kind: Management
metadata:
  name: kcm
spec:
  providers:
  - name: k0smotron
  - name: cluster-api-provider-aws
  - name: cluster-api-provider-azure
  - name: cluster-api-provider-vsphere
  - name: projectsveltos
  release: kcm-0-1-0

There are two options to override the default management configuration of KCM:

1. Update the Management object after the KCM installation using kubectl:

   kubectl --kubeconfig <path-to-management-kubeconfig> edit management

2. Deploy KCM skipping the default Management object creation and provide your own Management configuration:

   • Create management.yaml file and configure core components and providers. See Management API.

   • Specify --create-management=false controller argument and install KCM:

   If installing using helm add the following parameter to the helm install command:

   --set="controller.createManagement=false"

   • Create kcm Management object after KCM installation:

   kubectl --kubeconfig <path-to-management-kubeconfig> create -f management.yaml

To create a ClusterDeployment:

1. Create Credential object with all credentials required.

   See Credential system docs for more information regarding this object.

2. Select the ClusterTemplate you want to use for the deployment. To list all available templates, run:

export KUBECONFIG=<path-to-management-kubeconfig>

kubectl get clustertemplate -n kcm-system

If you want to deploy hosted control plane template, make sure to check additional notes on Hosted control plane in k0rdent Docs, see Documentation.

2. Create the file with the ClusterDeployment configuration:

apiVersion: k0rdent.mirantis.com/v1alpha1
kind: ClusterDeployment
metadata:
  name: <cluster-name>
  namespace: <cluster-namespace>
spec:
  template: <template-name>
  credential: <credential-name>
  dryRun: <true/false>
  config:
    <cluster-configuration>

3. Create the ClusterDeployment object:

kubectl create -f clusterdeployment.yaml

4. Check the status of the newly created ClusterDeployment object:

kubectl -n <clusterdeployment-namespace> get ClusterDeployment <clusterdeployment-name> -o=yaml

5. Wait for infrastructure to be provisioned and the cluster to be deployed (the provisioning starts only when spec.dryRun is disabled):

kubectl -n <clusterdeployment-namespace> get cluster <clusterdeployment-name> -o=yaml

6. Retrieve the kubeconfig of your cluster deployment:

kubectl get secret -n kcm-system <clusterdeployment-name>-kubeconfig -o=jsonpath={.data.value} | base64 -d > kubeconfig

KCM ClusterDeployment supports two modes: with and without (default) dryRun.

If no configuration (spec.config) provided, the ClusterDeployment object will be populated with defaults (default configuration can be found in the corresponding Template status) and automatically marked as dryRun.

Here is an example of the ClusterDeployment object with default configuration:

apiVersion: k0rdent.mirantis.com/v1alpha1
kind: ClusterDeployment
metadata:
  name: <cluster-name>
  namespace: <cluster-namespace>
spec:
  config:
    clusterNetwork:
      pods:
        cidrBlocks:
        - 10.244.0.0/16
      services:
        cidrBlocks:
        - 10.96.0.0/12
    controlPlane:
      iamInstanceProfile: control-plane.cluster-api-provider-aws.sigs.k8s.io
      instanceType: ""
    controlPlaneNumber: 3
    k0s:
      version: v1.27.2+k0s.0
    publicIP: false
    region: ""
    sshKeyName: ""
    worker:
      iamInstanceProfile: nodes.cluster-api-provider-aws.sigs.k8s.io
      instanceType: ""
    workersNumber: 2
  template: aws-standalone-cp-0-1-0
  credential: aws-credential
  dryRun: true

After you adjust your configuration and ensure that it passes validation (TemplateReady condition from status.conditions), remove the spec.dryRun flag to proceed with the deployment.

Here is an example of a ClusterDeployment object that passed the validation:

apiVersion: k0rdent.mirantis.com/v1alpha1
kind: ClusterDeployment
metadata:
  name: aws-standalone
  namespace: kcm-system
spec:
  template: aws-standalone-cp-0-1-0
  credential: aws-credential
  config:
    region: us-east-2
    publicIP: true
    controlPlaneNumber: 1
    workersNumber: 1
    controlPlane:
      instanceType: t3.small
    worker:
      instanceType: t3.small
  status:
    conditions:
    - lastTransitionTime: "2024-07-22T09:25:49Z"
      message: Template is valid
      reason: Succeeded
      status: "True"
      type: TemplateReady
    - lastTransitionTime: "2024-07-22T09:25:49Z"
      message: Helm chart is valid
      reason: Succeeded
      status: "True"
      type: HelmChartReady
    - lastTransitionTime: "2024-07-22T09:25:49Z"
      message: ClusterDeployment is ready
      reason: Succeeded
      status: "True"
      type: Ready
    observedGeneration: 1

1. Remove the Management object:

kubectl delete management.k0rdent kcm

2. Remove the kcm Helm release:

helm uninstall kcm -n kcm-system

3. Remove the kcm-system namespace:

kubectl delete ns kcm-system