Google Kubernetes Engine (GKE)

Provision, import, and operate Google Kubernetes Engine (GKE) clusters with Ankra - Standard and Autopilot, live pricing, node pools, upgrades, GitOps, and AI-assisted operations.

Google Kubernetes Engine (GKE) is Google Cloud’s managed Kubernetes service. Google runs the control plane; Ankra provisions the cluster, then manages everything on top - node pools, Kubernetes upgrades, addons, stacks, GitOps, and AI-assisted operations.

This page covers GKE specifically. For the shared concepts across all managed providers - live options, discovery/import, and day-2 API patterns - see Managed Kubernetes.

Why GKE with Ankra

• Standard or Autopilot - run node pools you size yourself, or let Google manage nodes with Autopilot.
• Release channels - pin the cluster to the RAPID, REGULAR, or STABLE channel.
• Live pricing and options - locations, Kubernetes versions, and machine types are fetched live with your service account, including GPU and spot capacity where available.

Prerequisites

A GCP service account key stored in Ankra, with the Kubernetes Engine Admin (roles/container.admin) and Service Account User (roles/iam.serviceAccountUser) roles, and the Kubernetes Engine API enabled on the project. To create VPC networks from the cluster wizard, the service account additionally needs Compute Network Admin (roles/compute.networkAdmin) for the compute.networks.create permission. See Google Cloud (GCP) Credentials.

GitOps is optional: connect a repository at creation and Ankra commits the cluster’s stack definitions to Git.

Creating a GKE cluster

Portal

1. Create Cluster

   Go to Clusters → Create Cluster and pick Google Cloud’s Cloud Managed action.

2. Credential & Location

   Select the GCP credential and a location - a zone (europe-west1-b) or a region (europe-west1). Locations load live from Google Cloud.

   Then pick the VPC Network the cluster attaches to. Ankra lists the project’s networks live; if the project has a default network it is preselected. Create new network provisions a fresh VPC from Ankra without leaving the wizard - see VPC networking.

3. Mode & Node Pools

   Choose Standard and define worker pools (name, machine type with live pricing, count, labels, autoscaling), or choose Autopilot and let Google manage the nodes.

4. Kubernetes

   Pick a Kubernetes version or keep the default, and select a release channel.

5. GitOps (optional) & Create

   Optionally connect a Git repository, then create. Ankra runs GKE preflight checks, provisions the cluster, retrieves the kubeconfig, and installs the Ankra Agent.

CLI

ankra cluster managed create \
  --provider gke \
  --name my-gke-cluster \
  --credential-id <gcp-credential-id> \
  --location europe-west1 \
  --node-pool-name workers \
  --node-pool-size e2-standard-4 \
  --node-pool-count 2

The release channel, Autopilot mode, and node-pool autoscaling bounds are set from the portal or the API.

API

curl -X POST https://platform.ankra.app/org/clusters/managed/gke \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "my-gke-cluster",
    "credential_id": "<gcp-credential-id>",
    "location": "europe-west1",
    "kubernetes_version": "1.32.3",
    "node_pools": [
      {
        "name": "workers",
        "size": "e2-standard-4",
        "count": 2,
        "autoscaling": {"enabled": true, "min_count": 2, "max_count": 5}
      }
    ],
    "gke": {"release_channel": "REGULAR", "mode": "standard"}
  }'

GKE options

Field	Description
gke.release_channel	RAPID, REGULAR, or STABLE
gke.mode	standard (you manage node pools) or autopilot (Google manages nodes)
gke.network	The VPC network the cluster attaches to, by name. Omitted, GKE assumes a network literally named default - which many projects don’t have, so pick one explicitly
gke.subnetwork	Optional subnetwork override. For custom-mode networks Ankra resolves the subnetwork in the cluster’s region automatically when this is omitted

Autopilot clusters ignore node-pool sizing - Google manages the nodes automatically. Cluster names are lowercase letters, numbers, and hyphens, and must start with a letter.

VPC networking

Every GKE cluster attaches to a VPC network. The wizard lists the project’s networks live with your credential and preselects default when the project has one; projects without a default network must pick one explicitly - otherwise the GKE API fails the create with “no network named default”. The subnetwork for the cluster’s region is resolved automatically (auto-mode networks have one per region; for custom-mode networks Ankra picks the region’s subnetwork).

Don’t have a usable network? Create new network in the wizard provisions one from Ankra: the VPC is created in auto subnet mode - Google creates a subnetwork in every region, exactly like the built-in default network - so the cluster can attach in any location with no further subnet setup. The same flow is available on the API:

curl -X POST https://platform.ankra.app/api/v1/clusters/managed/gke/networks \
  -H "Authorization: Bearer $ANKRA_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"credential_id": "<gcp-credential-id>", "name": "my-vpc"}'

Creating a network requires the compute.networks.create permission, which is not part of the base GKE provisioning roles. Grant Compute Network Admin (roles/compute.networkAdmin) to the service account - see Google Cloud (GCP) Credentials. Without it, the create fails with Required 'compute.networks.create' permission; attaching to an existing network keeps working.

Day-2 operations

Node pools, upgrades, and deletion work from the CLI, portal, or API. The CLI examples use --provider gke:

# scale a pool to a fixed count
ankra cluster managed node-pool scale <cluster_id> workers --provider gke --count 5

# upgrade the cluster
ankra cluster managed upgrade <cluster_id> --provider gke --version 1.32.3

# delete the cluster
ankra cluster managed delete <cluster_id> --provider gke

Autoscaling bounds are set from the portal or the PATCH .../node-pools/{name} endpoint. See Managed Kubernetes - day-2 operations for the shared mechanics.

Importing an existing GKE cluster

Already running GKE? Discover clusters in your project and adopt them into Ankra without touching them. Discovery and import run from the portal or API - see Managed Kubernetes - importing existing clusters.

Related

• Managed Kubernetes overview
• Google Cloud (GCP) Credentials
• Cluster settings