Refer to these instructions to provision a TKG cluster based on a custom ClusterClass. Note that these instructions are specific to vSphere 8 U1 environments.

Prerequisites

The procedure for provisioning a TKG cluster based on a custom ClusterClass is available starting with the vSphere 8 U1 release. If you are using vSphere 8 U2, see v1beta1 Example: Cluster Based on a Custom ClusterClass (vSphere 8 U2 and Later Workflow).

Adhere to the following prerequisites.
  • vSphere 8 U1 environment
  • Workload Management enabled
  • Supervisor configured
  • Ubuntu client with Kubernetes CLI Tools for vSphere installed
Attention: Custom ClusterClass is an experimental Kubernetes feature per the upstream Cluster API documentation. Due to the range of customizations available with custom ClusterClass, VMware cannot test or validate all possible customizations. Customers are responsible for testing, validating, and troubleshooting their custom ClusterClass clusters. Customers can open support tickets regarding their custom ClusterClass clusters, however, VMware support is limited to a best effort basis only and cannot guarantee resolution to every issue opened for custom ClusterClass clusters. Customers should be aware of these risks before deploying custom ClusterClass clusters in production environments.

Part 1: Create the Custom ClusterClass

The first part involves creating a custom ClusterClass by cloning the default ClusterClass which is named tanzukubernetescluster.
  1. Create a vSphere Namespace named custom-ns.
  2. Log in to Supervisor.
  3. Switch context to the vSphere Namespace named custom-ns.
  4. Get the default ClusterClass.
    kubectl get clusterclass tanzukubernetescluster -o json
  5. Create a custom ClusterClass named custom-cc by cloning the default ClusterClass.
    kubectl get clusterclass tanzukubernetescluster -o json | jq '.metadata.name="custom-cc"' | kubectl apply -f -
    Expected result:
    clusterclass.cluster.x-k8s.io/custom-cc created
  6. Get the custom ClusterClass.
    kubectl get clusterclass custom-cc -o json

    If necessary you can use less to view the custom ClusterClass.

    kubectl get clusterclass custom-cc -o json | less
    Note: Issue command "q" to exit less.

Part 2: Create Required Supervisor Objects to Provision the TKG Cluster

The next part is to create the Supervisor objects required for the initial deployment of a custom TKG cluster using the custom ClusterClass.
Note: We use the cluster name "ccc-cluster" by default. If you are using a different cluster name, you will need to change it appropriate fields.
  1. Create the issuer for the self-signed extensions certificate.
    #self-signed-extensions-issuer.yaml
    apiVersion: cert-manager.io/v1
    kind: Issuer
    metadata:
      name: self-signed-extensions-issuer
    spec:
      selfSigned: {}
    kubectl apply -f self-signed-extensions-issuer.yaml -n custom-ns
    Expected result:
    issuer.cert-manager.io/self-signed-extensions-issuer created
  2. Create the secret for the extensions CA certificate.
    #extensions-ca-certificate.yaml
    apiVersion: cert-manager.io/v1
    kind: Certificate
    metadata:
      name: ccc-cluster-extensions-ca
    spec:
      commonName: kubernetes-extensions
      duration: 87600h0m0s
      isCA: true
      issuerRef:
        kind: Issuer
        name: self-signed-extensions-issuer
      secretName: ccc-cluster-extensions-ca
      usages:
      - digital signature
      - cert sign
      - crl sign
    kubectl apply -f extensions-ca-certificate.yaml -n custom-ns
    Expected result:
    certificate.cert-manager.io/ccc-cluster-extensions-ca created
  3. Create the issuer for the extensions CA certificate.
    #extensions-ca-issuer.yaml
    apiVersion: cert-manager.io/v1
    kind: Issuer
    metadata:
      name: ccc-cluster-extensions-ca-issuer
    spec:
      ca:
        secretName: ccc-cluster-extensions-ca
    kubectl apply -f extensions-ca-issuer.yaml -n custom-ns
    Expected result:
    issuer.cert-manager.io/ccc-cluster-extensions-ca-issuer created
  4. Create the secrete for the auth service certificate.
    #auth-svc-cert.yaml
    apiVersion: cert-manager.io/v1
    kind: Certificate
    metadata:
      name: ccc-cluster-auth-svc-cert
    spec:
      commonName: authsvc
      dnsNames:
      - authsvc
      - localhost
      - 127.0.0.1
      duration: 87600h0m0s
      issuerRef:
        kind: Issuer
        name: ccc-cluster-extensions-ca-issuer
      secretName: ccc-cluster-auth-svc-cert
      usages:
      - server auth
      - digital signature
    kubectl apply -f auth-svc-cert.yaml -n custom-ns
    Expected result:
    certificate.cert-manager.io/ccc-cluster-auth-svc-cert created
  5. Verify the creation of the issuers and certificates.
    kubectl get issuers -n custom-ns
    NAME                               READY   AGE
    ccc-cluster-extensions-ca-issuer   True    2m57s
    self-signed-extensions-issuer      True    14m
    
    kubectl get certs -n custom-ns
    NAME                        READY   SECRET                      AGE
    ccc-cluster-auth-svc-cert   True    ccc-cluster-auth-svc-cert   34s
    ccc-cluster-extensions-ca   True    ccc-cluster-extensions-ca   5m
    

Part 3: Create a TKG Cluster Based on the Custom ClusterClass

You use the Cluster v1beta1 API to create a Cluster based on a ClusterClass. A v1beta1 Cluster based on a custom ClusterClass requires the following minimum set of variables.
Variable Description
vmClass See Using VM Classes with TKG Service Clusters.
storageClass See Configure Persistent Storage for the vSphere Namespace.
ntp NTP server used to enable Supervisor.
extensionCert Auto-generated after the "extension CA certificate" was created in the previous section.
clusterEncryptionConfigYaml section below will walk through the process of getting this file
  1. Create the encryption secret.
    #encryption-secret.yaml
    apiVersion: v1
    data:
      key: all3dzZpODFmRmh6MVlJbUtQQktuN2ViQzREbDBQRHlxVk8yYXRxTW9QQT0=
    kind: Secret
    metadata:
      name: ccc-cluster-encryption
    type: Opaque
    kubectl apply -f encryption-secret.yaml -n custom-ns
    Expected result:
    secret/ccc-cluster-encryption created
  2. Gather the NTP server from Supervisor.
    kubectl -n vmware-system-vmop get configmap vmoperator-network-config -o jsonpath={.data.ntpservers}
  3. Construct the cluster-with-ccc.yaml manifest to provision the cluster.
    #cluster-with-ccc.yaml
    apiVersion: cluster.x-k8s.io/v1beta1
    kind: Cluster
    metadata:
      name: ccc-cluster
    spec:
      clusterNetwork:
        pods:
          cidrBlocks:
          - 193.0.0.0/16
        serviceDomain: managedcluster1.local
        services:
          cidrBlocks:
          - 198.201.0.0/16
      topology:
        class: custom-cc
        version: v1.26.5---vmware.2-fips.1-tkg.1
        controlPlane:
          metadata: {}
          replicas: 3
        workers:
          machineDeployments:
            - class: node-pool
              metadata: { }
              name: node-pool-workers
              replicas: 3
        variables:
        - name: vmClass
          value: guaranteed-medium
        - name: storageClass
          value: tkg-storage-profile
        - name: ntp
          value: time.acme.com
        - name: extensionCert
          value:
            contentSecret:
              key: tls.crt
              name: ccc-cluster-extensions-ca
        - name: clusterEncryptionConfigYaml
          value: LS0tCm...Ht9Cg==
    In the cluster manifest, verify or update the following fields:
    Parameter Description
    metadata.name Name of the v1beta1 Cluster.
    spec.topology.class Name of the custom ClusterClass.
    spec.topology.version Tanzu Kubernetes release version
    spec.topology.variables.storageClass.value StoragePolicy attached to the vSphere Namespace where the cluster will be provisioned
    spec.topology.variables.ntp.value NTP server address
    spec.topology.variables.extensionCert.value.contentSecret.name Verify
    spec.topology.variables.clusterEncryptionConfigYaml.value Populate the with the data.key value from the ClusterEncryptionConfig secret
  4. Create the cluster based on the custom ClusterClass.
    kubectl apply -f cluster-with-ccc.yaml -n custom-ns
    Expected result:
    cluster.cluster.x-k8s.io/ccc-cluster created

    Using the vSphere Client, verify that the cluster is created.

  5. Log in to the TKG cluster.
    kubectl vsphere login --server=xxx.xxx.xxx.xxx --vsphere-username USERNAME@vsphere.local --tanzu-kubernetes-cluster-name ccc-cluster --tanzu-kubernetes-cluster-namespace custom-ns

Part 4: Create Required Supervisor Objects to Manage the TKG Cluster

Once the cluster with CCC is applied, various controllers will try to provision it; however, underlying infrastructure resources still require additional objects to be bootstrapped properly.
Parameter Value
Authentication Authentication values need to be gathered and updated into a file named values.yaml
Base64 encoded The values.yaml file will be encoded into a base64 string
guest-cluster-auth-service-data-values.yaml This string will be added to the guest-cluster-auth-service-data-values.yaml file downloaded from CCC_config_yamls.tar.gz prior to applying the file
GuestClusterAuthSvcDataValues secret Finally, the Guest Cluster Bootstrap must be modified to reference the newly created GuestClusterAuthSvcDataValues secret.
  1. Switch context to the vSphere Namespace where the cluster is provisioned.
    kubectl config use-context custom-ns
  2. Get the authServicePublicKeys value.
    kubectl -n vmware-system-capw get configmap vc-public-keys -o jsonpath="{.data.vsphere\.local\.json}"
    Copy the result to a text file named values.yaml.
    authServicePublicKeys: '{"issuer_url":"https://...SShrDw=="]}]}}'
  3. Get the cluster UID to update the authServicePublicKeys.
    kubectl get cluster -n custom-ns ccc-cluster -o yaml | grep uid
  4. In the authServicePublicKeys section of the values.yaml file, append the cluster UID to the "client_id" value.

    Syntax: vmware-tes:vc:vns:k8s:clusterUID

    For example:
    vmware-tes:vc:vns:k8s:7d95b50b-4fd4-4642-82a3-5dbfe87f499c
  5. Get the certificate value (replace ccc-cluster with the chosen cluster name).
    kubectl -n custom-ns get secret ccc-cluster-auth-svc-cert -o jsonpath="{.data.tls\.crt}" | base64 -d
  6. Add the certificate to the values.yaml.

    Add the certificate contents beneath the authServicePublicKeys section.

    Note: The certificate must be indented 4 spaces to avoid failure.
    For example:
    authServicePublicKeys: '{"issuer_url":"https://...SShrDw=="]}]}}'
    ceritificate: |
        -----BEGIN CERTIFICATE-----
        MIIDPTCCAiWgAwIBAgIQMibGSjeuJelQoPxCof/+xzANBgkqhkiG9w0BAQsFADAg
        ...
        sESk/RDTB1UAvi8PD3zcbEKZuRxuo4IAJqFFbAabwULhjUo0UwT+dIJo1gLf5/ep
        VoIRJS7j6VT98WbKyZp5B4I=
        -----END CERTIFICATE-----
  7. Get the privateKey value.
    kubectl -n custom-ns get secret ccc-cluster-auth-svc-cert -o jsonpath="{.data.tls\.key}"
  8. Verify your values.yaml file.
    authServicePublicKeys: '{"issuer_url":"https://10.197.79.141/openidconnect/vsphere.local","client_id":"vmware-tes:vc:vns:k8s:7d95...499c",...SShrDw=="]}]}}'
    certificate: |
        -----BEGIN CERTIFICATE-----
        MIIDPTCCAiWgAwIBAgIQWQyXAQDRMhgrGre8ysVN0DANBgkqhkiG9w0BAQsFADAg
        ...
        uJSBP49sF0nKz5nf7w+BdYE=
        -----END CERTIFICATE-----
    privateKey: LS0tLS1CRUdJTi...VktLS0tLQo=
    
  9. Hash the values.yaml file with base64-encoding to gather output for the guest-cluster-auth-service-data-values.yaml file.
    base64 -i values.yaml -w 0
  10. Create the guest-cluster-auth-service-data-values.yaml file.
    Here is a template for the secret.
    apiVersion: v1
    data:
      values.yaml: YXV0a...ExRbz0K
    kind: Secret
    metadata:
      labels:
        tkg.tanzu.vmware.com/cluster-name: ccc-cluster
        tkg.tanzu.vmware.com/package-name: guest-cluster-auth-service.tanzu.vmware.com.1.3.0+tkg.2-vmware
      name: ccc-cluster-guest-cluster-auth-service-data-values
    type: Opaque
    Refer to the following table to populate the expected secret values.
    Parameter Value
    data.values.yaml

    Base64-encoded string of values.yaml

    metadata.labels.cluster-name

    Name of the cluster, such as ccc-cluster

    metadata.labels.package-name

    guest-cluster-auth-service.tanzu.vmware.com.version

    To get this value, run the command kubectl get tkr v1.26.5---vmware.2-fips.1-tkg.1 -o yaml

    Change the TKR version depending on the version you are using

    metadata.name

    Name of the cluster, such as ccc-cluster

  11. Create the guest-cluster-auth-service-data-values.yaml secret.
    kubectl apply -f guest-cluster-auth-service-data-values.yaml -n custom-ns
  12. Edit Cluster Bootstrap to reference the secret.
    kubectl edit clusterbootstrap ccc-cluster -n custom-ns
  13. Add the following lines beneath the line guest-cluster-auth-service.tanzu.vmware.com.version:.
    valuesFrom:
      secretRef: ccc-cluster-guest-cluster-auth-service-data-values
    For example:
    spec:
      additionalPackages:
      - refName: guest-cluster-auth-service.tanzu.vmware.com.1.3.0+tkg.2-vmware
        valuesFrom:
          secretRef: ccc-cluster-guest-cluster-auth-service-data-values
  14. Save and quit to apply the clusterbootstrap modifications.

Part 5: Configure Pod Security

If you are using TKR versions 1.25 and later, configure pod security for the vSphere Namespace named custom-ns. See Configure PSA for TKR 1.25 and Later.

If you are using TKR versions 1.24 and prior, pods in the cluster require binding to pod security policy. To apply the required resource objects on the cluster level, use the following process.
  1. Gather the TKG cluster kubeconfig.
    kubectl -n custom-ns get secret ccc-cluster-kubeconfig -o jsonpath="{.data.value}" | base64 -d > ccc-cluster-kubeconfig
  2. Create the psp.yaml file.
    apiVersion: policy/v1beta1
    kind: PodSecurityPolicy
    metadata:
      name: tanzu-system-kapp-ctrl-restricted
    spec:
      privileged: false
      allowPrivilegeEscalation: false
      requiredDropCapabilities:
        - ALL
      volumes:
        - configMap
        - emptyDir
        - projected
        - secret
        - downwardAPI
        - persistentVolumeClaim
      hostNetwork: false
      hostIPC: false
      hostPID: false
      runAsUser:
        rule: MustRunAsNonRoot
      seLinux:
        rule: RunAsAny
      supplementalGroups:
        rule: MustRunAs
        ranges:
          - min: 1
            max: 65535
      fsGroup:
        rule: MustRunAs
        ranges:
          - min: 1
            max: 65535
      readOnlyRootFilesystem: false
  3. Apply the pod security policy.
    KUBECONFIG=ccc-cluster-kubeconfig kubectl apply -f psp.yaml
  4. Log in to the TKG cluster.
    kubectl vsphere login --server=10.197.154.66 --vsphere-username administrator@vsphere.local --insecure-skip-tls-verify --tanzu-kubernetes-cluster-name ccc-cluster --tanzu-kubernetes-cluster-namespace custom-ns
  5. List the namespaces.
    KUBECONFIG=ccc-cluster-kubeconfig kubectl get ns -A
    NAME                           STATUS   AGE
    default                        Active   13d
    kube-node-lease                Active   13d
    kube-public                    Active   13d
    kube-system                    Active   13d
    secretgen-controller           Active   13d
    tkg-system                     Active   13d
    vmware-system-antrea           Active   13d
    vmware-system-cloud-provider   Active   13d
    vmware-system-csi              Active   13d
    vmware-system-tkg              Active   13d
    

Part 6: Synchronize vSphere SSO Roles with the Custom TKG Cluster

Rolebinding for vCenter Single Sign-On users built in the vSphere Namespaces must be synchronized from Supervisor to the TKG cluster for developers to manage the cluster workloads.

This process requires exporting the existing rolebinding list from Supervisor, gathering rolebindings that have "edit" role and creating the file sync-cluster-edit-rolebinding.yaml and then applying to the TKG cluster using its KUBECONFIG.
  1. Gather existing rolebindings from Supervisor.
    kubectl get rolebinding -n custom-ns  -o yaml
  2. From the returned list of rolebinding objects, identify ones with roleRef.name equal to "edit".
    For example:
    apiVersion: v1
    items:
    - apiVersion: rbac.authorization.k8s.io/v1
      kind: RoleBinding
      metadata:
        creationTimestamp: "2023-08-25T18:44:45Z"
        name: ccc-cluster-8lr5x-ccm
        namespace: custom-ns
        ownerReferences:
        - apiVersion: vmware.infrastructure.cluster.x-k8s.io/v1beta1
          blockOwnerDeletion: true
          controller: true
          kind: ProviderServiceAccount
          name: ccc-cluster-8lr5x-ccm
          uid: b5fb9f01-9a55-4f69-8673-fadc49012994
        resourceVersion: "108766602"
        uid: eb93efd4-ae56-4d9f-a745-d2782885e7fb
      roleRef:
        apiGroup: rbac.authorization.k8s.io
        kind: Role
        name: ccc-cluster-8lr5x-ccm
      subjects:
      - kind: ServiceAccount
        name: ccc-cluster-8lr5x-ccm
        namespace: custom-ns
    - apiVersion: rbac.authorization.k8s.io/v1
      kind: RoleBinding
      metadata:
        creationTimestamp: "2023-08-25T18:44:45Z"
        name: ccc-cluster-8lr5x-pvcsi
        namespace: custom-ns
        ownerReferences:
        - apiVersion: vmware.infrastructure.cluster.x-k8s.io/v1beta1
          blockOwnerDeletion: true
          controller: true
          kind: ProviderServiceAccount
          name: ccc-cluster-8lr5x-pvcsi
          uid: d9342f8f-13d2-496d-93cb-b24edfacb5c1
        resourceVersion: "108766608"
        uid: fd1820c7-7993-4299-abb7-bb67fb17f1fd
      roleRef:
        apiGroup: rbac.authorization.k8s.io
        kind: Role
        name: ccc-cluster-8lr5x-pvcsi
      subjects:
      - kind: ServiceAccount
        name: ccc-cluster-8lr5x-pvcsi
        namespace: custom-ns
    - apiVersion: rbac.authorization.k8s.io/v1
      kind: RoleBinding
      metadata:
        creationTimestamp: "2023-08-25T16:58:06Z"
        labels:
          managedBy: vSphere
        name: wcp:custom-ns:group:vsphere.local:administrators
        namespace: custom-ns
        resourceVersion: "108714148"
        uid: d74a98c7-e7da-4d71-b1d5-deb60492d429
      roleRef:
        apiGroup: rbac.authorization.k8s.io
        kind: ClusterRole
        name: edit
      subjects:
      - apiGroup: rbac.authorization.k8s.io
        kind: Group
        name: sso:Administrators@vsphere.local
    - apiVersion: rbac.authorization.k8s.io/v1
      kind: RoleBinding
      metadata:
        creationTimestamp: "2023-08-25T16:58:21Z"
        labels:
          managedBy: vSphere
        name: wcp:custom-ns:user:vsphere.local:administrator
        namespace: custom-ns
        resourceVersion: "108714283"
        uid: 07f7dbba-2670-4100-a59b-c09e4b2edd6b
      roleRef:
        apiGroup: rbac.authorization.k8s.io
        kind: ClusterRole
        name: edit
      subjects:
      - apiGroup: rbac.authorization.k8s.io
        kind: User
        name: sso:Administrator@vsphere.local
    kind: List
    metadata:
      resourceVersion: ""
    
  3. Create a file named sync-cluster-edit-rolebinding.yaml to add any extra rolebindings other than the default administrator@vsphere.local. For example:
    apiVersion: rbac.authorization.k8s.io/v1
    kind: ClusterRoleBinding
    metadata:
      labels:
        run.tanzu.vmware.com/vmware-system-synced-from-supervisor: "yes"
      name: vmware-system-auth-sync-wcp:custom-ns:group:vsphere.local:administrators
    roleRef:
      apiGroup: rbac.authorization.k8s.io
      kind: ClusterRole
      name: cluster-admin
    subjects:
    - apiGroup: rbac.authorization.k8s.io
      kind: Group
      name: sso:Administrators@vsphere.local
    apiVersion: rbac.authorization.k8s.io/v1
    kind: ClusterRoleBinding
    metadata:
      labels:
        run.tanzu.vmware.com/vmware-system-synced-from-supervisor: "yes"
      name: vmware-system-auth-sync-wcp:custom-ns:group:SSODOMAIN.COM:testuser
    roleRef:
      apiGroup: rbac.authorization.k8s.io
      kind: ClusterRole
      name: cluster-admin
    subjects:
    - apiGroup: rbac.authorization.k8s.io
      kind: Group
      name: sso:testuser@SSODOMAIN.COM
    Note: In the metadata.name field, the user role is prepended with vmware-system-auth-sync- for all users. The metadata.name and subjects.name entries will require modification for all non-default roles.
  4. Apply the sync-cluster-edit-rolebinding.yaml configuration to synchronize rolebindings.
    KUBECONFIG=ccc-cluster-kubeconfig kubectl apply -f sync-cluster-edit-rolebinding.yaml