Create a Custom ClusterClass

This topic describes how to create your own custom ClusterClass resource for a Tanzu Kubernetes Grid (TKG) standalone management cluster, use it to create class-based workload clusters, and work with clusters that are based on the custom ClusterClass.

To base a cluster on a custom ClusterClass, you set its spec.topology.class to the custom ClusterClass name.

These procedures do not apply to TKG with a vSphere with Tanzu Supervisor.

Caution

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. To create workload clusters using the default ClusterClass resource, follow the procedures in Steps for Cluster Deployment.

Requirements

To create a custom ClusterClass you need imgpkg, the Tanzu CLI, and kubectl installed locally. If you are using ytt templates as a source for your customizations, you also need ytt installed. For information about how to download and install ytt and imgpkg, see Install the Carvel Tools.

Options: Templates vs. New Object Spec

To create a custom ClusterClass, VMware recommends starting with the existing, default ClusterClass manifest or YTT templates described under Create a Base ClusterClass Manifest. When a new version of the default ClusterClass object is published, for example with a new version of TKG, you can then apply the overlay to the new version in order to implement the same customizations.

In v1.x versions of TKG, standalone management clusters used YTT overlays stored on the bootstrap machine to customize the clusters that they created. You can use these same overlays as sources for a shared custom ClusterClass in TKG v2.

The procedures in this topic describe this method of creating a custom ClusterClass object.

To write an entirely new ClusterClass object without using an existing template, follow the procedure in Writing a ClusterClass in the Cluster API documentation.

ClusterClass variables and patches

For cluster values that are neither part of the Cluster object’s topology nor supplied by its underlying objects, a ClusterClass can define them in its variables and patches blocks:

  • variables blocks define variables that are settable within the Cluster object itself, like workerMachineType with name, type, default setting, etc., following the OpenAPI Specification v3 standard. For example, if a ClusterClass variables block specifies a required variable named imageRepository of type string and with a default value of k8s.gcr.io, then every Cluster object that the ClusterClass creates will include variable.imageRepository, defaulted and settable as specified in the Cluster object spec.

  • patches blocks apply Sprig functions to variable values in the ClusterClass and in the object templates that it references to derive new values, and then use jsonPatch rules to read and set those values. In this way, patches provide a general-purpose mechanism that enables settings within a Cluster object to determine other settings within the cluster itself and its underlying objects.

Create a Base ClusterClass Manifest

You can create a base ClusterClass manifest based on the default ClusterClass manifest or on YTT templates that Tanzu Kubernetes Grid provides. Any custom clusters that you create should be based on this base ClusterClass manifest. There are 3 steps to the process:

  1. Obtain the default ClusterClass manifest or YTT templates for your target platform.
  2. Use the default ClusterClass manifest or YTT templates to generate a customized base ClusterClass manifest, that includes information about your infrastructure.
  3. Use this base ClusterClass manifest as the template from which to create your customized clusters.

There are three methods in which you can create a base ClusterClass manifest for your clusters.

  • Method 1: Obtain the Base Manifest From a Management Cluster Directly. This is the simplest method. A default base manifest is automatically generated when you deploy a management cluster. This base manifest includes the infrastructure configuration that you provided when you deployed the management cluster. You can use this as a base ClusterClass manifest directly.
  • Method 2: Obtain YTT Templates from the Tanzu CLI (Advanced). When you install the Tanzu CLI, YTT templates are installed on your machine. To create a base manifest from these YTT templates, you must create a YTT overlay that provides your infrastructure configuration.
  • Method 3: Obtain YTT Templates from the TKG Image Registry (Advanced). You can also obtain the YTT templates from the TKG image registry. To create a base manifest from these YTT templates, you must create a YTT overlay that provides your infrastructure configuration.
Important

Methods 2 and 3 are for advanced users who need to satisfy the following use-cases:

  • You want to generate custom ClusterClass definitions for your CI system, without deploying a standalone management cluster.
  • You want workload clusters to use different infrastructure than management clusters.

Method 1: Obtain the Base Manifest From a Management Cluster Directly

In Tanzu Kubernetes Grid 2.3.0 and later, after you deploy a management cluster, you can find the default ClusterClass manifest in the ~/.config/tanzu/tkg/clusterclassconfigs folder.

To see the manifest for the target platforms on which you have deployed a management cluster, run the following command:

tree ~/.config/tanzu/tkg/clusterclassconfigs/

For example, if you deployed a management cluster to vSphere, you will see the following YAML file.

.config/tanzu/tkg/clusterclassconfigs/
└── tkg-vsphere-default-v1.1.1.yaml
 
1 directory, 1 file

The generated manifest contains information about your target platform that has been extracted from your management cluster deployment. You can use this as the base manifest for custom ClusterClass creation directly. For the next steps, see Customize Your Base ClusterClass Manifest.

Method 2: Obtain YTT Templates from the Tanzu CLI

After you install the Tanzu CLI v0.90.1 or later, but before you deploy a management cluster, you can find the YTT templates for the default ClusterClass in the ~/.config/tanzu/tkg/providers/infrastructure-<provider name>/<provider version>/cconly folder. You can use these templates to create a base manifest for custom ClusterClass creation.

  1. To find the templates, run the appropriate command for your target platform.

    vSphere
    tree ~/.config/tanzu/tkg/providers/infrastructure-vsphere/v1.7.0/cconly
    
    You will see the following YAML files.
    .config/tanzu/tkg/providers/infrastructure-vsphere/v1.7.0/cconly
    ├── base.yaml
    ├── overlay-kube-apiserver-admission.yaml
    └── overlay.yaml
    
    1 directory, 3 files
    
    AWS
    tree ~/.config/tanzu/tkg/providers/infrastructure-aws/v2.1.3/cconly
    
    You will see the following YAML files.
    .config/tanzu/tkg/providers/infrastructure-aws/v2.1.3/cconly
    ├── base.yaml
    ├── overlay-kube-apiserver-admission.yaml
    └── overlay.yaml
    
    Azure
    tree ~/.config/tanzu/tkg/providers/infrastructure-azure/v1.9.2/cconly/
    
    You will see the following YAML files.
    .config/tanzu/tkg/providers/infrastructure-azure/v1.9.2/cconly/
    ├── base.yaml
    ├── overlay-kube-apiserver-admission.yaml
    └── overlay.yaml
    
  2. Create a YTT data-value file named user_input.yaml with the following content.

    #@data/values
    
    #@overlay/match-child-defaults missing_ok=True
    ---
    
  3. Add configuration values appropriate to your target platform to user_input.yaml.

    You can use the configuration values that you used when you deployed a management cluster deployment. For example, a user_input.yaml file for vSphere will resemble the following:

    #@data/values
    
    #@overlay/match-child-defaults missing_ok=True
    ---
    ENABLE_MHC: true
    ENABLE_MHC_CONTROL_PLANE: true
    ENABLE_MHC_WORKER_NODE: true
    MHC_UNKNOWN_STATUS_TIMEOUT: 5m
    MHC_FALSE_STATUS_TIMEOUT: 12m
    MHC_MAX_UNHEALTHY_CONTROL_PLANE: 100%
    MHC_MAX_UNHEALTHY_WORKER_NODE: 100%
    NODE_STARTUP_TIMEOUT: 20m
    VSPHERE_TLS_THUMBPRINT: 0A:B4:8F:2E:E4:34:82:90:D5:6A:F8:77:8C:8C:51:24:D2:49:3B:E8
    VSPHERE_DATACENTER: /dc0
    VSPHERE_DATASTORE: /dc0/datastore/sharedVmfs-0
    VSPHERE_FOLDER: /dc0/vm
    VSPHERE_NETWORK: /dc0/network/VM Network
    VSPHERE_RESOURCE_POOL: /dc0/host/cluster0/Resources
    VSPHERE_SERVER: xxx.xxx.xxx.xxx
    VSPHERE_USERNAME: [email protected]
    VSPHERE_CONTROL_PLANE_DISK_GIB: "20"
    VSPHERE_CONTROL_PLANE_MEM_MIB: "8192"
    VSPHERE_CONTROL_PLANE_NUM_CPUS: "4" VSPHERE_WORKER_DISK_GIB: "20"
    VSPHERE_WORKER_MEM_MIB: "8192"
    VSPHERE_WORKER_NUM_CPUS: "4"
    VSPHERE_CLUSTER_CLASS_VERSION: "v1.1.1"       # change it to the version in TKG BOM
    NAMESPACE: "tkg-system"                       # DO NOT change it
    
  4. Use ytt to generate the base ClusterClass manifest for your target platform.

    vSphere
    ytt -f ~/.config/tanzu/tkg/providers/infrastructure-vsphere/v1.7.0/cconly -f ~/.config/tanzu/tkg/providers/config_default.yaml -f user_input.yaml
    
    AWS
    ytt -f ~/.config/tanzu/tkg/providers/infrastructure-aws/v2.1.3/cconly -f ~/.config/tanzu/tkg/providers/config_default.yaml -f user_input.yaml
    
    Azure
    ytt -f ~/.config/tanzu/tkg/providers/infrastructure-azure/v1.9.2/cconly -f ~/.config/tanzu/tkg/providers/config_default.yaml -f user_input.yaml
    

Now that you have generated a base manifest for your infrastructure, for the next steps, see Customize Your Base ClusterClass Manifest.

Method 3: Obtain YTT Templates from the TKG Image Registry

The YTT templates for the default ClusterClass are also available for download from the TKG image repository. You can use these templates to create a base manifest for custom ClusterClass creation.

  1. Pull the YTT templates from the providerTemplate image in the official TKG registry.

    For TKG v2.4.1, the providerTemplate image tag is v0.31.1. You can find the tag version in the TKG BOM file by searching for providerTemplateImage.

    imgpkg pull -i projects.registry.vmware.com/tkg/tanzu_core/provider/provider-templates:v0.31.1 -o providers
    

    You will see output similar to the following:

    Pulling image 'projects.registry.vmware.com/tkg/tanzu_core/provider/provider-templates@sha256:b210d26c610800f5da4b3aa55bfbc8ffae9275fa2c4073a2b1332e2045a6e1da'
    Extracting layer 'sha256:3ba336232c0e616b2b6c8f263593497c5a059a645f4c6137ea0eb658f4a8538a' (1/1)
    
    Succeeded
    

    The YAML template files are downloaded into a providers folder.

  2. Create a YTT data-value file named user_input.yaml with the following content.

    #@data/values
    
    #@overlay/match-child-defaults missing_ok=True
    ---
    
  3. Add configuration values appropriate to your target platform to user_input.yaml.

    You can use the configuration values that you used when you deployed a management cluster deployment. For example, a user_input.yaml file for vSphere will resemble the following:

    #@data/values
    
    #@overlay/match-child-defaults missing_ok=True
    ---
    ENABLE_MHC: true
    ENABLE_MHC_CONTROL_PLANE: true
    ENABLE_MHC_WORKER_NODE: true
    MHC_UNKNOWN_STATUS_TIMEOUT: 5m
    MHC_FALSE_STATUS_TIMEOUT: 12m
    MHC_MAX_UNHEALTHY_CONTROL_PLANE: 100%
    MHC_MAX_UNHEALTHY_WORKER_NODE: 100%
    NODE_STARTUP_TIMEOUT: 20m
    VSPHERE_TLS_THUMBPRINT: 0A:B4:8F:2E:E4:34:82:90:D5:6A:F8:77:8C:8C:51:24:D2:49:3B:E8
    VSPHERE_DATACENTER: /dc0
    VSPHERE_DATASTORE: /dc0/datastore/sharedVmfs-0
    VSPHERE_FOLDER: /dc0/vm
    VSPHERE_NETWORK: /dc0/network/VM Network
    VSPHERE_RESOURCE_POOL: /dc0/host/cluster0/Resources
    VSPHERE_SERVER: xxx.xxx.xxx.xxx
    VSPHERE_USERNAME: [email protected]
    VSPHERE_CONTROL_PLANE_DISK_GIB: "20"
    VSPHERE_CONTROL_PLANE_MEM_MIB: "8192"
    VSPHERE_CONTROL_PLANE_NUM_CPUS: "4" VSPHERE_WORKER_DISK_GIB: "20"
    VSPHERE_WORKER_MEM_MIB: "8192"
    VSPHERE_WORKER_NUM_CPUS: "4"
    VSPHERE_CLUSTER_CLASS_VERSION: "v1.1.1"       # change it to the version in TKG BOM
    NAMESPACE: "tkg-system"                       # DO NOT change it
    
  4. Use ytt to generate the base ClusterClass manifest for your target platform.

    vSphere
    ytt -f providers/infrastructure-vsphere/v1.7.0/cconly -f providers/config_default.yaml -f user_input.yaml
    
    AWS
    ytt -f providers/infrastructure-aws/v2.1.3/cconly -f providers/config_default.yaml -f user_input.yaml
    
    Azure
    ytt -f ~/.config/tanzu/tkg/providers/infrastructure-azure/v1.9.2/cconly -f ~/.config/tanzu/tkg/providers/config_default.yaml -f user_input.yaml
    

Now that you have generated a base manifest for your infrastructure, for the next steps, see Customize Your Base ClusterClass Manifest.

Customize Your Base ClusterClass Manifest

To customize your ClusterClass manifest, you create ytt overlay files alongside the manifest. The following example shows how to modify a Linux kernel parameter in the ClusterClass definition.

  1. Create a custom folder structured as follows:

    tree custom
    custom
    |-- overlays
        |-- filter.yaml
        |-- kernels.yaml
    
  2. Edit custom/overlays/kernels.yaml.

    For example, add nfConntrackMax as a variable and define a patch for it that adds its value to the kernel parameter net.netfilter.nf_conntrack_max for control plane nodes.

    This overlay appends a command to the field preKubeadmCommands, to write the configuration to sysctl.conf. To make the setting take effect you append the command sysctl -p to apply the change. Default ClusterClass definitions are immutable, so this overlay also changes the name of your custom ClusterClass and all of its templates by adding -extended.

    This overlay also removes the run.tanzu.vmware.com/tkg-managed label if it is present, to ensure that this ClusterClass is treated as a custom object, not a default and a managed one, during workflows such as upgrade.

    cat custom/overlays/kernels.yaml
    #@ load("@ytt:overlay", "overlay")
    
    #@overlay/match by=overlay.subset({"kind":"ClusterClass"})
    ---
    apiVersion: cluster.x-k8s.io/v1beta1
    kind: ClusterClass
    metadata:
      name: tkg-vsphere-default-v1.1.1-extended
      labels:
        #@overlay/match missing_ok=True
        #@overlay/remove
        run.tanzu.vmware.com/tkg-managed: ""
    spec:
      variables:
      - name: nfConntrackMax
        required: false
        schema:
          openAPIV3Schema:
            type: string
      patches:
      - name: nfConntrackMax
        enabledIf: '{{ not (empty .nfConntrackMax) }}'
        definitions:
          - selector:
              apiVersion: controlplane.cluster.x-k8s.io/v1beta1
              kind: KubeadmControlPlaneTemplate
              matchResources:
                controlPlane: true
            jsonPatches:
              - op: add
                path: /spec/template/spec/kubeadmConfigSpec/preKubeadmCommands/-
                valueFrom:
                  template: echo "net.netfilter.nf_conntrack_max={{ .nfConntrackMax }}" >> /etc/sysctl.conf
              - op: add
                path: /spec/template/spec/kubeadmConfigSpec/preKubeadmCommands/-
                value: sysctl -p
    
  3. Remove resources that you do not want to change.

    In this example, all templates are intended to be shared between the custom and default ClusterClass so they are all removed. You can also create a custom template based on the default template in the same way, in which case make sure that kind is not excluded.

    cat custom/overlays/filter.yaml
    #@ load("@ytt:overlay", "overlay")
    
    #@overlay/match by=overlay.not_op(overlay.subset({"kind": "ClusterClass"})),expects="0+"
    ---
    #@overlay/remove
    
  4. Use the default ClusterClass manifest to generate the base ClusterClass.

    ytt -f tkg-vsphere-default-v1.1.1.yaml -f custom/overlays/filter.yaml > default_cc.yaml
    
  5. Generate the custom ClusterClass.

    ytt -f tkg-vsphere-default-v1.1.1.yaml -f custom/ > custom_cc.yaml
    
  6. (Optional) Check the difference between the default ClusterClass and your custom one, to confirm that the changes have been applied.

    diff default_cc.yaml custom_cc.yaml
    

    You should see output similar to the following:

    4c4
    <   name: tkg-vsphere-default-v1.1.1
    ---
    >   name: tkg-vsphere-default-v1.1.1-extended
    638a639,643
    >   - name: nfConntrackMax
    >     required: false
    >     schema:
    >       openAPIV3Schema:
    >         type: string
    2607a2613,2628
    >   - name: nfConntrackMax
    >     enabledIf: '{{ not (empty .nfConntrackMax) }}'
    >     definitions:
    >     - selector:
    >         apiVersion: controlplane.cluster.x-k8s.io/v1beta1
    >         kind: KubeadmControlPlaneTemplate
    >         matchResources:
    >           controlPlane: true
    >       jsonPatches:
    >       - op: add
    >         path: /spec/template/spec/kubeadmConfigSpec/preKubeadmCommands/-
    >         valueFrom:
    >           template: echo "net.netfilter.nf_conntrack_max={{ .nfConntrackMax }}" >> /etc/sysctl.conf
    >       - op: add
    >         path: /spec/template/spec/kubeadmConfigSpec/preKubeadmCommands/-
    >         value: sysctl -p
    

In this example, you can see that -extended has been added to the cluster name.

Install the Custom ClusterClass in the Management Cluster

To enable your management cluster to use your custom ClusterClass, install it by applying the new manifest.

  1. Apply the ClusterClass manifest.

    kubectl apply -f custom_cc.yaml
    

    You should see the following output.

    clusterclass.cluster.x-k8s.io/tkg-vsphere-default-v1.1.1-extended created
    
  2. Check if the custom ClusterClass has propagated to the default namespace, for example:

    kubectl get clusterclass
    

    You should see the following output.

    NAME                                  AGE
    tkg-vsphere-default-v1.1.1            2d23h
    tkg-vsphere-default-v1.1.1-extended   11s
    

Generate the Custom Workload Cluster Manifest

After creating your custom ClusterClass, you can use it to create a new workload cluster that includes your customization.

  1. Run tanzu cluster create with the --dry-run option to generate a cluster manifest from a standard cluster configuration file.

    tanzu cluster create --file workload-1.yaml --dry-run > default_cluster.yaml
    
  2. Create a ytt overlay or edit the cluster manifest directly.

    The recommended option is to create a ytt overlay, for example cluster_overlay.yaml, to do the following:

    • Replace the topology.class value with the name of your custom ClusterClass
    • Add your custom variables to the variables block, with a default value

    As with modifying ClusterClass object specs, using an overlay as follows lets you automatically apply the changes to new objects whenever there is a new upstream cluster release.

    #@ load("@ytt:overlay", "overlay")
    
    #@overlay/match by=overlay.subset({"kind":"Cluster"})
    ---
    apiVersion: cluster.x-k8s.io/v1beta1
    kind: Cluster
    spec:
      topology:
        class: tkg-vsphere-default-v1.1.1-extended
        variables:
        - name: nfConntrackMax
          value: "1048576
    
  3. Generate the custom_cluster.yaml manifest:

    ytt -f default_cluster.yaml -f cluster_overlay.yaml > custom_cluster.yaml
    
  4. (Optional) As with the ClusterClass above, you can run diff to compare your custom-class cluster manifest with a default class-based cluster, for example:

    diff custom_cluster.yaml default_cluster.yaml
    

    You should see output similar to the following:

    <     class: tkg-vsphere-default-v1.1.1
    ---
    >     class: tkg-vsphere-default-v1.1.1-extended
    142a143,144
    >     - name: nfConntrackMax
    >       value: "1048576"
    

Create a Custom Cluster

Create a custom workload cluster based on the custom manifest generated above as follows.

  1. Create the cluster.

    tanzu cluster create -f custom_cluster.yaml
    
  2. Check created object properties.

    For example with the kernel modification above, retrieve the KubeadmControlPlane object and confirm that the kernel configuration is set:

    kubectl get kcp workload-1-jgwd9 -o yaml
    

    You should see output similar to the following:

    apiVersion: controlplane.cluster.x-k8s.io/v1beta1
    kind: KubeadmControlPlane
    ...
        preKubeadmCommands:
        - hostname "{{ ds.meta_data.hostname }}"
        - echo "::1         ipv6-localhost ipv6-loopback" >/etc/hosts
        - echo "127.0.0.1   localhost" >>/etc/hosts
        - echo "127.0.0.1   {{ ds.meta_data.hostname }}" >>/etc/hosts
        - echo "{{ ds.meta_data.hostname }}" >/etc/hostname
        - echo "net.netfilter.nf_conntrack_max=1048576" >> /etc/sysctl.conf
        - sysctl -p
    ...
    
  3. Log in to a control plane node and confirm that its sysctl.conf is modified:

    capv@workload-1-jgwd9:~$ sudo cat /etc/sysctl.conf
    ...
    net.ipv6.neigh.default.gc_thresh3=16384
    fs.file-max=9223372036854775807
    net.netfilter.nf_conntrack_max=1048576
    

Upgrade Custom Clusters

If you created custom clusters with a previous release, you can upgrade them to the latest TKG release.

Important

If your custom ClusterClass definition is based on a default ClusterClass, make sure the label run.tanzu.vmware.com/tkg-managed is not present in its metadata.labels field. Otherwise, upgrading the cluster may replace your custom ClusterClass with the default.

Prepare to Upgrade Clusters

Before you upgrade clusters, there are preparatory steps to perform.

  1. Before you upgrade the management cluster, create test clusters with the version of the custom manifest that you created for the previous release.

    For example, create a custom cluster named test-upgrade.

  2. Get information about the ClusterClass versions available with your TKG 2.2 management cluster.

    kubectl get clusterclass
    

    If you created the custom clusters with TKG v2.2.0, the ClusterClass version should be 1.0.0. For example:

    NAME                                  AGE
    tkg-vsphere-default-extended-v1.0.0   21m
    tkg-vsphere-default-v1.0.0            10d
    
  3. Get the information about the ClusterClass versions running in your test-upgrade cluster.

    kubectl get cluster test-upgrade -o jsonpath='{.spec.topology.class}'
    

    The output should be tkg-vsphere-default-extended-v1.0.0.

  4. Follow the instructions in Upgrade Standalone Management Clusters to upgrade the management cluster to TKG 2.4.

  5. Get information about the ClusterClass version available after upgrading the management cluster to v.2.4.

    kubectl get cluster mgmt-cluster-name -n tkg-system -o jsonpath='{.spec.topology.class}'
    

    The output should be tkg-vsphere-default-v1.1.1 if the management cluster is running on vSphere.

  6. Follow the procedures in Create a Base ClusterClass Manifest and Customize Your Base ClusterClass Manifest to create a new version of your ClusterClass manifest.
    • For example name the new custom ClusterClass tkg-vsphere-default-v1.1.1-extended and include the same custom variables as in the old version tkg-vsphere-default-extended-v1.0.0.
    • Name the new overlay custom_cc.yaml.
  7. Install the new custom ClusterClass in the management cluster.

    kubectl apply -f custom_cc.yaml
    
  8. Get information about the ClusterClass versions that are now available with your management cluster.

    kubectl get clusterclass
    

    Both the older versions and the newer versions should be displayed.

    NAME                                  AGE
    tkg-vsphere-default-extended-v1.0.0   61m
    tkg-vsphere-default-v1.0.0            10d
    tkg-vsphere-default-v1.1.1            25m
    tkg-vsphere-default-v1.1.1-extended   15s
    

Perform the Upgrade

After you have performed the preparatory steps, you can test the upgrade on the test cluster before you upgrade your production clusters.

  1. Rebase the cluster test-upgrade from the old version of the custom ClusterClass to the new one.

    kubectl patch cluster test-upgrade --type merge -p '{"spec": {"topology": {"class": "tkg-vsphere-default-v1.1.1-extended"}}}'   
    

    The output should be cluster.cluster.x-k8s.io/test-upgrade patched.

  2. Get the information about the ClusterClass version now running in your test-upgrade cluster.

    kubectl get cluster test-upgrade -o jsonpath='{.spec.topology.class}'
    

    The output should be tkg-vsphere-default-v1.1.1-extended. Previously it was tkg-vsphere-default-extended-v1.0.0.

  3. Wait for several minutes and run kubectl get cluster until you see that UpdatesAvailable is updated to true.

    kubectl get cluster test-upgrade -o yaml
    

    When ready, a message similar to the following should appear in the output:

    ...
    status:
      conditions:
    ...
      - lastTransitionTime: "2023-06-19T09:59:21Z"
        message: '[v1.25.9+vmware.1-tkg.1-20230609 v1.26.4+vmware.1-tkg.1-20230609]'
        status: "True"
        type: UpdatesAvailable
      controlPlaneReady: true
      infrastructureReady: true
      observedGeneration: 5
      phase: Provisioned
    
  4. Upgrade the test-upgrade cluster.

    tanzu cluster upgrade test-upgrade    
    
  5. Check created object properties.

    For example with the kernel modification described in the example above, retrieve the KubeadmControlPlane object and confirm that the kernel configuration is set:

    kubectl get kcp test-upgrade-nsc6d -o yaml
    

    You should see output similar to the following:

    apiVersion: controlplane.cluster.x-k8s.io/v1beta1
    kind: KubeadmControlPlane
    ...
        preKubeadmCommands:
        - hostname "{{ ds.meta_data.hostname }}"
        - echo "::1         ipv6-localhost ipv6-loopback" >/etc/hosts
        - echo "127.0.0.1   localhost" >>/etc/hosts
        - echo "127.0.0.1   {{ ds.meta_data.hostname }}" >>/etc/hosts
        - echo "{{ ds.meta_data.hostname }}" >/etc/hostname
        - sed -i 's|".*/pause|"projects-stg.registry.vmware.com/tkg/pause|' /etc/containerd/config.toml
        - systemctl restart containerd
        - echo "net.netfilter.nf_conntrack_max=1048576" >> /etc/sysctl.conf
        - sysctl -p  
    

Upgrade Production Clusters

If the test upgrade was successful, repeat the steps from Perform the Upgrade on your production clusters.

Note

If you encounter any errors during the upgrade, you can roll back by rebasing the cluster from the new version of the custom ClusterClass to the old version.

Rebase a Default Cluster to Use a Custom ClusterClass

If you have created clusters with the default ClusterClass and want to update them to use a custom ClusterClass, use kubectl to edit the Cluster object:

  • Change spec.topology.class to the name of your custom ClassClass manifest.
  • Modify spec.topology.variables to append your custom variables.

Rebase a Custom Cluster to Use the Default ClusterClass

If you want to revert to a new version of the default ClusterClass:

  • Change spec.topology.class to the new version of the default ClusterClass.
  • Modify spec.topology.variables to remove your custom variables. You might need to add new variables that are defined in the new version of default ClusterClass.
check-circle-line exclamation-circle-line close-line
Scroll to top icon