Create and Manage GCP IaaS Cluster

Palette supports creating and managing Kubernetes clusters using Google Kubernetes Engine (GKE). This section guides you to create a Kubernetes cluster that is deployed to GKE and that Palette manages.


Ensure the following requirements are met before you attempt to deploy a cluster to GCP.

  • Access to a GCP cloud account.
  • Palette creates compute, network, and storage resources while provisioning Kubernetes clusters. Ensure there is sufficient capacity in the preferred GCP region to create the following resources:
    • Virtual Private Cloud (VPC) Network
    • Static External IP Address
    • Network Interfaces
    • Cloud NAT
    • Cloud Load Balancing
    • Persistent Disks
    • Cloud Router

Deploy a GKE Cluster

  1. Log in to Palette and ensure you are in the correct project scope.
  1. Navigate to the left Main Menu and click on Clusters.
  1. Click on Add New Cluster.
  1. A prompt displays to either deploy or import a new cluster. Click on Deploy New Cluster.
  1. Select GCP and click on Start GCP Configuration.
  1. Populate the wizard page with the cluster name, description, and tags. Tags assigned to a cluster are propagated to the VMs deployed to the computing environments.

  2. Select a GCP account, and click on Next.

  1. Select the Managed Kubernetes row and select one of your GKE cluster profiles. Click on Next.
  1. Review and customize pack parameters as desired. By default, parameters for all packs are set with values defined in the cluster profile. Click on Next to continue.
  1. Fill out the following parameters, and click on Next when you are done.

    ProjectThe project to which the cluster belongs.
    RegionChoose the desired GCP region in which to deploy the cluster.
  1. The Node configuration page is where you can specify the availability zones (AZ), instance types, disk size, and the number of nodes. Configure the worker node pool.

    You can add new worker pools to customize specific worker nodes to run specialized workloads. For example, the default worker pool may be configured with the c2.standarnd-4 instance types for general-purpose workloads. You can configure another worker pool with instance type g2-standard-4 to run GPU workloads.

  1. An optional taint label can be applied to a node pool during the cluster creation. You can edit the taint label on existing clusters. Review the Node Pool management page to learn more. Toggle the Taint button to create a label.
  1. Enable or disable node pool taints. If tainting is enabled, then you need to provide values for the following parameters.

    KeyCustom key for the taint.
    ValueCustom value for the taint key.
    EffectChoose the preferred pod scheduling effect from the drop-down Menu. Review the Effect Table below for more details.

    Effect Table

    NoScheduleA pod that cannot tolerate the node taint and should not be scheduled to the node.
    PreferNoScheduleThe system will avoid placing a non-tolerant pod to the tainted node but is not guaranteed.
    NoExecuteNew pods will not be scheduled on the node, and existing pods on the node will be evicted if they do not tolerate the taint.
  2. Click on Next after configuring the node pool.

  1. The Settings page is where you can configure the patching schedule, security scans, backup settings, and set up Role Based Access Control (RBAC). Review cluster settings and make changes if needed. Click on Validate.
  1. Review the settings summary and click on Finish Configuration to deploy the cluster. Be aware that provisioning GKE clusters can take 15 - 30 minutes depending on the cluster profile and the node pool configuration.

You can monitor cluster deployment progress on the cluster details page.


You can validate that your cluster is up and available by reviewing the cluster details page.

  1. Log in to Palette.
  1. Navigate to the left Main Menu and click on Clusters.
  1. The Clusters page lists the available clusters that Palette manages. Select your cluster to view its details page.
  1. From the cluster details page, verify the Cluster Status field displays Running.