gcloud-container

name	gcloud-container
description	Google Kubernetes Engine (GKE) via gcloud (`gcloud container`). Deploy and manage clusters of machines for running containers — ai, attached, aws, azure, bare-metal, binauthz, clusters, fleet.

gcloud container — Google Kubernetes Engine (GKE)

Overview

gcloud container manages Google Kubernetes Engine (GKE) — Google Cloud's managed Kubernetes service — and related container tooling. Use it to create and tear down clusters, manage node pools, fetch kubectl credentials, upgrade Kubernetes versions, and administer fleets, Anthos/attached clusters, and Container Registry images. GKE runs in two modes: Autopilot (Google manages nodes, scaling, and security; created with clusters create-auto) and Standard (you manage node pools and machine types; created with clusters create).

Reach for this group whenever you provision GKE infrastructure or wire kubectl up to a cluster. Day-to-day workload operations (deployments, pods, services) are done with kubectl after running clusters get-credentials.

Quick reference — common workflows

Enable the API

gcloud services enable container.googleapis.com

Create a Standard cluster and connect kubectl

# Check the default and available Kubernetes versions for a location
gcloud container get-server-config --location=us-central1

# Create a zonal Standard cluster (defaults to 3 nodes)
gcloud container clusters create my-cluster \
    --location=us-central1-a \
    --release-channel=regular \
    --num-nodes=3 \
    --machine-type=e2-standard-4 \
    --disk-type=pd-balanced \
    --disk-size=100

# Write credentials to ~/.kube/config so kubectl targets this cluster
gcloud container clusters get-credentials my-cluster \
    --location=us-central1-a

Create an Autopilot cluster

# Autopilot manages nodes automatically — use create-auto, not create
gcloud container clusters create-auto my-autopilot-cluster \
    --location=us-central1 \
    --release-channel=regular

gcloud container clusters get-credentials my-autopilot-cluster \
    --location=us-central1

List and inspect clusters

gcloud container clusters list

gcloud container clusters describe my-cluster \
    --location=us-central1-a

# Auto-upgrade status, upgrade targets, end-of-support timelines
gcloud container clusters get-upgrade-info my-cluster \
    --location=us-central1-a

Manage node pools (Standard clusters)

# Add a spot-VM node pool with autoscaling and autorepair
gcloud container node-pools create high-mem-pool \
    --cluster=my-cluster \
    --location=us-central1-a \
    --machine-type=n2-highmem-4 \
    --num-nodes=2 \
    --spot \
    --enable-autoscaling \
    --min-nodes=1 \
    --max-nodes=10 \
    --enable-autorepair

gcloud container node-pools list \
    --cluster=my-cluster \
    --location=us-central1-a

# Manually resize a specific node pool
gcloud container clusters resize my-cluster \
    --node-pool=high-mem-pool \
    --num-nodes=5 \
    --location=us-central1-a

gcloud container node-pools delete high-mem-pool \
    --cluster=my-cluster \
    --location=us-central1-a

Upgrade a cluster

# Upgrade the control plane (master) to the default version
gcloud container clusters upgrade my-cluster \
    --master \
    --location=us-central1-a

# Upgrade a node pool to a specific version (node pools can't upgrade with the master)
gcloud container clusters upgrade my-cluster \
    --node-pool=default-pool \
    --cluster-version=1.30.5-gke.1014001 \
    --location=us-central1-a

# Roll back a canceled or failed node-pool upgrade
gcloud container node-pools rollback default-pool \
    --cluster=my-cluster \
    --location=us-central1-a

Update cluster settings (autoscaling, monitoring)

# Enable node autoscaling on the default node pool
gcloud container clusters update my-cluster \
    --enable-autoscaling \
    --min-nodes=1 \
    --max-nodes=10 \
    --location=us-central1-a

# Switch the autoscaling profile and enable managed Prometheus
gcloud container clusters update my-cluster \
    --autoscaling-profile=optimize-utilization \
    --enable-managed-prometheus \
    --location=us-central1-a

Command groups

Group	Reference file	Commands	Description
container ai	ai.md	9	manage AI related workloads for GKE
container attached	attached.md	12	manage Attached clusters for running containers
container aws	aws.md	17	deploy and manage clusters of machines on AWS for running containers
container azure	azure.md	21	deploy and manage clusters of machines on Azure for running containers
container bare-metal	bare-metal.md	23	deploy and manage Anthos clusters on bare metal for running containers
container binauthz	binauthz.md	21	manage attestations for Binary Authorization on Google Cloud Platform
container clusters	clusters.md	11	deploy and teardown GKE clusters (create, create-auto, upgrade, resize, get-credentials)
container fleet	fleet.md	112	centrally manage features and services on all your Kubernetes clusters with fleet
container hub	hub.md	112	alias for container fleet — centrally manage fleet features and services
container images	images.md	7	list and manipulate Google Container Registry images
container node-pools	node-pools.md	8	create, update, and delete GKE node pools
container operations	operations.md	4	get and list operations for GKE clusters
container subnets	subnets.md	1	manage subnets to be used by GKE clusters
container vmware	vmware.md	22	deploy and manage Anthos clusters on VMware for running containers

Top-level command: gcloud container get-server-config (see _commands.md) returns the default and valid Kubernetes/image versions for a location. The full one-line index of all 381 GA commands is in index.md.

Common flags & tips

• Location is required for most cluster/node-pool commands. Prefer --location=LOCATION (accepts either a zone like us-central1-a or a region like us-central1) over the older --zone/--region flags. A zonal value creates a single-zone cluster; a regional value creates a regional cluster with the control plane replicated across zones.
• Autopilot vs Standard: clusters create-auto produces an Autopilot cluster (no node-pool management, no --num-nodes/--machine-type); clusters create produces a Standard cluster where you size node pools yourself. --release-channel (rapid, regular, stable, extended) controls automatic version management on both.
• get-credentials is the bridge to kubectl. It writes endpoint and auth data to ~/.kube/config (override with $KUBECONFIG). Use --internal-ip or --dns-endpoint for private/DNS-based control-plane access.
• Upgrades are split: --master upgrades the control plane; --node-pool=NAME upgrades nodes. They cannot be upgraded in the same command. Omitting --cluster-version defaults node pools to the current master version and the master to the channel default.
• Node pool naming: node-pool commands take the pool name positionally and require --cluster=CLUSTER. Resizing is done via clusters resize --node-pool=..., not a node-pool subcommand.
• Useful formatting/filtering:

  # Tabular view of clusters with version and node count
  gcloud container clusters list \
      --format="table(name,location,currentMasterVersion,currentNodeCount,status)"
  
  # Only RUNNING clusters
  gcloud container clusters list --filter="status=RUNNING"
  
  # Just the endpoint of one cluster
  gcloud container clusters describe my-cluster \
      --location=us-central1-a --format="value(endpoint)"
  

• container hub is an alias for container fleet; both expose the same 112 commands.

beta / alpha

• gcloud beta container clusters mirrors the GA surface but exposes additional/earlier flags. It adds the two-step staged control-plane upgrade, whose completion command is beta-only (not present in GA container clusters):

  gcloud beta container clusters upgrade CLUSTER_NAME \
      --master --location=LOCATION \
      --cluster-version=VERSION \
      --control-plane-soak-duration=SOAK_DURATION
  gcloud beta container clusters complete-control-plane-upgrade CLUSTER_NAME \
      --location=LOCATION
  

• gcloud alpha container clusters offers experimental features (e.g. create-auto, check-autopilot-compatibility, complete-control-plane-upgrade) at the alpha tier; alpha clusters are not covered by the GKE SLA and should not run production workloads.
• gcloud container ai profiles ... (GA) manages AI/ML workload profiles — accelerator profiles, model servers, and manifests — for GKE.

Official documentation

• GKE product documentation home — entry point for all GKE concepts, how-tos, and reference material.
• Create Standard zonal/regional clusters — how-to for clusters create.
• Create Autopilot clusters — how-to for clusters create-auto.
• Autopilot overview — Autopilot mode and comparison with Standard.
• Cluster architecture — control plane, node pools, and mode differences.
• Node pools how-to — create, describe, resize, update, and delete node pools.
• Upgrading a cluster — control plane and node-pool version upgrades.
• GKE IAM — roles required for cluster management (e.g. roles/container.admin, roles/container.clusterAdmin).
• gcloud container reference — full command/flag reference for this group.