GKE cluster creation (Autopilot vs Standard) in GCP - CLI Comparison

Introduction

Creating a Kubernetes cluster on Google Cloud lets you run and manage containerized apps easily. You can choose Autopilot mode for hands-off management or Standard mode for full control over your cluster's resources.

When you want Google to manage the infrastructure and scaling automatically without manual intervention.

When you need full control over node configuration and want to customize your cluster setup.

When you want to quickly deploy a cluster without worrying about node management.

When you want to optimize costs by only paying for the resources your pods use.

When you want to run workloads that require specific node customizations or special hardware.

Commands

This command creates a GKE Autopilot cluster named 'example-autopilot-cluster' in the 'us-central1' region. Autopilot mode means Google manages the nodes and scaling for you.

Terminal

gcloud container clusters create-auto example-autopilot-cluster --region us-central1

Expected Output

Creating cluster example-autopilot-cluster in us-central1...



Created [https://container.googleapis.com/v1/projects/my-project/locations/us-central1/clusters/example-autopilot-cluster].



To inspect the contents of your cluster, go to: https://console.cloud.google.com/kubernetes/list?project=my-project



kubeconfig entry generated for example-autopilot-cluster.



NAME                        LOCATION      MASTER_VERSION  MASTER_IP       MACHINE_TYPE  NODE_VERSION  NUM_NODES  STATUS

example-autopilot-cluster   us-central1   1.26.3-gke.100  35.202.123.45   Autopilot    1.26.3-gke.100  3          RUNNING

→

--region - Specifies the region where the cluster will be created.

This command creates a GKE Standard cluster named 'example-standard-cluster' in the 'us-central1-a' zone with 3 nodes of type 'e2-medium'. You manage the nodes and scaling yourself.

Terminal

gcloud container clusters create example-standard-cluster --zone us-central1-a --num-nodes 3 --machine-type e2-medium

Expected Output

Creating cluster example-standard-cluster in us-central1-a...



Created [https://container.googleapis.com/v1/projects/my-project/zones/us-central1-a/clusters/example-standard-cluster].



To inspect the contents of your cluster, go to: https://console.cloud.google.com/kubernetes/list?project=my-project



kubeconfig entry generated for example-standard-cluster.



NAME                     LOCATION        MASTER_VERSION  MASTER_IP       MACHINE_TYPE  NODE_VERSION  NUM_NODES  STATUS

example-standard-cluster  us-central1-a  1.26.3-gke.100  34.68.234.56    e2-medium    1.26.3-gke.100  3          RUNNING

→

--zone - Specifies the zone where the cluster will be created.

→

--num-nodes - Sets the initial number of nodes in the cluster.

→

--machine-type - Defines the type of virtual machine for each node.

This command lists the nodes in the current Kubernetes cluster to verify the cluster is running and nodes are ready.

Terminal

kubectl get nodes

Expected Output

NAME                                       STATUS   ROLES    AGE   VERSION

example-standard-cluster-default-pool-1   Ready    <none>   2m    v1.26.3-gke.100

Key Concept

If you remember nothing else from this pattern, remember: Autopilot mode lets Google manage your cluster nodes automatically, while Standard mode gives you full control over node setup and scaling.

Common Mistakes

Trying to use node customization flags with Autopilot clusters.

Autopilot mode does not allow manual node configuration; it manages nodes automatically.

Use Standard clusters if you need to customize nodes or control scaling.

Not specifying region or zone correctly when creating clusters.

The cluster creation will fail or create in an unexpected location.

Always specify --region for Autopilot and --zone for Standard clusters explicitly.

Assuming Autopilot clusters have visible nodes with kubectl get nodes.

Autopilot nodes are managed by Google and may not appear as standard nodes.

Use kubectl get pods and other commands to check workload status in Autopilot clusters.

Summary

Use 'gcloud container clusters create-auto' to create an Autopilot cluster where Google manages nodes.

Use 'gcloud container clusters create' with zone, node count, and machine type to create a Standard cluster with manual node control.

Verify cluster nodes with 'kubectl get nodes' to ensure your cluster is running and ready.

Practice

1. What is the main difference between a GKE Autopilot cluster and a Standard cluster?

easy

A. Autopilot manages the infrastructure automatically, while Standard requires manual node management.

B. Standard clusters automatically scale nodes, Autopilot does not.

C. Autopilot clusters only support Windows nodes, Standard supports Linux nodes.

D. Standard clusters are serverless, Autopilot clusters require manual server setup.

Solution

1. Step 1: Understand Autopilot cluster management

   Autopilot clusters handle node provisioning, scaling, and maintenance automatically for you.

2. Step 2: Understand Standard cluster management

   Standard clusters require you to create and manage nodes manually, including scaling and updates.

3. Final Answer:

   Autopilot manages the infrastructure automatically, while Standard requires manual node management. -> Option A

4. Quick Check:

   Autopilot = automatic node management [OK]

Hint: Autopilot means Google manages nodes for you [OK]

Common Mistakes:

• Thinking Standard clusters auto-scale nodes by default
• Confusing Autopilot with serverless functions
• Believing Autopilot only supports Windows nodes

2. Which of the following is the correct gcloud command to create a GKE Autopilot cluster named my-autopilot-cluster?

easy

A. gcloud container clusters create my-autopilot-cluster --mode=autopilot

B. gcloud container clusters create my-autopilot-cluster --enable-autopilot

C. gcloud container clusters create-autopilot my-autopilot-cluster

D. gcloud container clusters create my-autopilot-cluster --autopilot=true

Solution

1. Step 1: Recall the gcloud syntax for Autopilot clusters

   The correct command uses gcloud container clusters create with the flag --enable-autopilot.

2. Step 2: Check other options for correctness

   Options A, B, and D use incorrect commands or flags not supported by gcloud.

3. Final Answer:

   gcloud container clusters create my-autopilot-cluster --enable-autopilot -> Option B

4. Quick Check:

   Use --enable-autopilot flag to create Autopilot clusters [OK]

Hint: Use --enable-autopilot flag with create command [OK]

Common Mistakes:

• Using non-existent commands like create-autopilot
• Using incorrect flags like --mode=autopilot
• Assuming --autopilot=true is valid

3. Consider this snippet creating a Standard GKE cluster:

gcloud container clusters create my-standard-cluster \
  --num-nodes=3 \
  --machine-type=e2-medium

What will be the number of nodes and machine type in the cluster?

medium

A. Cluster creation will fail due to missing flags

B. 1 node with e2-medium machine

C. 3 nodes with default machine type

D. 3 nodes with e2-medium machines

Solution

1. Step 1: Analyze the command flags

   The flag --num-nodes=3 sets the cluster to have 3 nodes, and --machine-type=e2-medium sets the machine type for each node.

2. Step 2: Confirm cluster creation behavior

   Standard clusters accept these flags and create nodes accordingly without error.

3. Final Answer:

   3 nodes with e2-medium machines -> Option D

4. Quick Check:

   --num-nodes=3 and --machine-type=e2-medium create 3 e2-medium nodes [OK]

Hint: Flags --num-nodes and --machine-type set nodes and size [OK]

Common Mistakes:

• Assuming default node count is 1 despite --num-nodes flag
• Thinking machine type defaults if specified
• Believing cluster creation fails without extra flags

4. You tried to create an Autopilot cluster with this command:

gcloud container clusters create my-autopilot-cluster --num-nodes=3 --enable-autopilot

But it failed. What is the likely cause?

medium

A. Autopilot clusters require a zone flag.

B. You must specify --machine-type with Autopilot clusters.

C. Autopilot clusters ignore --num-nodes and fail if specified.

D. The cluster name is invalid for Autopilot clusters.

Solution

1. Step 1: Understand Autopilot cluster restrictions

   Autopilot clusters manage node count automatically and do not accept the --num-nodes flag.

2. Step 2: Identify why the command fails

   Including --num-nodes causes the command to fail because it conflicts with Autopilot's automatic management.

3. Final Answer:

   Autopilot clusters ignore --num-nodes and fail if specified. -> Option C

4. Quick Check:

   Do not use --num-nodes with Autopilot clusters [OK]

Hint: Autopilot ignores --num-nodes; remove it to fix [OK]

Common Mistakes:

• Adding --machine-type to Autopilot clusters
• Assuming cluster name causes failure
• Forgetting to specify zone (not mandatory)

5. You want to create a GKE cluster that automatically manages node scaling and maintenance but also allows you to customize node pools with specific machine types. Which cluster type should you choose and why?

hard

A. Standard cluster, because it allows full control over node pools and manual scaling.

B. Autopilot cluster, because it manages nodes automatically and allows node pool customization.

C. Autopilot cluster, because it requires manual node management but auto-scales pods.

D. Standard cluster, because it automatically manages nodes and does not allow customization.

Solution

1. Step 1: Identify Autopilot cluster features

   Autopilot clusters manage nodes automatically but do not allow customizing node pools or machine types.

2. Step 2: Identify Standard cluster features

   Standard clusters require manual node management but allow full customization of node pools and machine types.

3. Step 3: Match requirements to cluster type

   You want automatic node management plus node pool customization, which is not possible with Autopilot. Standard clusters allow customization but require manual scaling.

4. Final Answer:

   Standard cluster, because it allows full control over node pools and manual scaling. -> Option A

5. Quick Check:

   Customization needs Standard cluster; Autopilot limits node control [OK]

Hint: Customize nodes? Choose Standard; Autopilot limits control [OK]

Common Mistakes:

• Thinking Autopilot allows node pool customization
• Assuming Autopilot auto-scales nodes and pods separately
• Believing Standard clusters auto-manage nodes