This is supported from TKG 1.4 version. When you deploy workload clusters using TKG 1.4, it create on node pool by default. You can validate that by running the following command.

# List the node pool in a workload cluster
# Ref command
$ tanzu cluster node-pool list 

$ tanzu cluster node-pool list tkg-wkld01
  NAME                       NAMESPACE  PHASE    REPLICAS  READY  UPDATED  UNAVAILABLE
  tkg-wkld01-md-0  default    Running  3        3     3       0

Pre-requirements for Creating a new node pool

Create a node pool definition file

Below is sample definition file for creating a new node pool. You can modify this as per your environment and requirement.

name: tkg-wkld02
replicas: 2
labels:
  key1: wkld02
  key2: app-workload
vsphere:
  memoryMiB: 4196
  diskGiB: 40
  numCPUs: 4
  datacenter: datacenter
  datastore: wkld02
  storagePolicyName: tkg-policy
  folder: tkg-folder
  resourcePool: tkg-pool
  network: net-seg2

Let’s understand the parameters:

• name: name of the node pool
• replicas: number of node to be created
• labels: you can specify the labels for better identification of nodes and later if you want to use in selector
• vsphere: vsphere specific parameters e.g. size of nodes when it will be created, datacenter name etc.

Note: If you don’s specify, these vsphere settings, they will be inherited from first node pool that was creating during installation time. There are more parameters available that you can specify within vsphere section, you can find there here. Ref: https://docs.vmware.com/en/VMware-Tanzu-Kubernetes-Grid/1.4/vmware-tanzu-kubernetes-grid-14/GUID-tanzu-k8s-clusters-node-pool.html

Creating node pool

Run below command to create the node pool.

# Add the new node pool
$ tanzu cluster node-pool set tkg-wkld01 -f nodepool

# List the node pool in a cluster 
$ tanzu cluster node-pool list tkg-wkld01
  NAME                       NAMESPACE  PHASE      REPLICAS  READY  UPDATED  UNAVAILABLE
  tkg-wkld01-md-0            default    Running    3         3      3       0
  tkg-wkld02                 default    ScalingUp  2         0      2        2

You can see the progress on vCenter console and also keep monitoring the node pool status.

# List the node pool again in a cluster
$ tanzu cluster node-pool list tkg-wkld01
  NAME                       NAMESPACE  PHASE      REPLICAS  READY  UPDATED  UNAVAILABLE
  tkg-wkld01-md-0            default    Running    3         3      3       0
  tkg-wkld02                 default    ScalingUp  2         0      2        2

You can also run the below command to see the node status

# Below are the newly added nodes in a cluster
$ kubectl get nodes
tkg-wkld02-78dfd5865c-8j2rb                  Ready                     6m22s   v1.21.2+vmware.1
tkg-wkld02-78dfd5865c-l2q9l                  Ready                     6m23s   v1.21.2+vmware.1

Deleting node pool

Now, I will delete the above node pool created just now. It’s very simple, you can run the below command to delete it.

# Delete the node pool
$  tanzu cluster node-pool delete tkg-wkld01 --name tkg-wkld02

• –name is the name of node pool object

Validate the node status from kubectl command.

# Validate if the nodes are deleted from list
$ kubectl get nodes

Updating node pool

Updating the existing node pool is simple and straight forward. e.g. below is the reference file for node pool. e.g.,

Scenario1:

let say you want to change the number of node in a cluster to 3, change the following.

name: tkg-wkld01
replicas: 3
labels:
  key1: wkld01
  key2: app-workload
vsphere:

Run the following command

# Update the existing node pool for number of node change
$ tanzu cluster node-pool set tkg-wkld01 -f nodepool

Scenario2:

let say you want to update the memory of nodes to 4GB

name: tkg-wkld01
replicas: 3
labels:
  key1: wkld01
  key2: app-workload
vsphere:
  memoryMiB: 4196

Run the following command

# Update the existing node pool for memory update
$ tanzu cluster node-pool set tkg-wkld01 -f nodepool

You will notice that the worker node will start getting created one by one and old one will be deleted. All the process runs in background and you can monitor that in vCenter.