Cluster Autoscaler concept in Kubernetes - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

How does Cluster Autoscaler decide to add nodes?

Cluster Autoscaler watches the pods and nodes in a Kubernetes cluster. When does it decide to add new nodes?

A When a node fails health checks repeatedly.

B When CPU usage on existing nodes exceeds 90%.

C When there are pods that cannot be scheduled due to insufficient resources on existing nodes.

D When the number of pods exceeds a fixed threshold regardless of resource availability.

💡 Hint

Think about what causes pods to wait for resources.

Cluster Autoscaler adds nodes only when pods cannot be scheduled because existing nodes lack enough CPU, memory, or other resources.

💻 Command Output

intermediate

Output of Cluster Autoscaler status command

What is the expected output of the command kubectl get deployment cluster-autoscaler -n kube-system -o jsonpath='{.status.availableReplicas}' if the Cluster Autoscaler is running with 1 replica?

A 1

B 0

C null

D Error: deployment not found

💡 Hint

Check the meaning of availableReplicas in deployment status.

The field availableReplicas shows how many pods of the deployment are ready and running. If Cluster Autoscaler is running with 1 pod, it shows 1.

❓ Configuration

advanced

Correct Cluster Autoscaler flag to ignore nodes with specific label

You want Cluster Autoscaler to ignore nodes labeled autoscaler=disabled. Which flag should you add to the Cluster Autoscaler deployment?

A --skip-nodes-with-label=autoscaler=disabled

B --ignore-nodes-with-label=autoscaler=disabled

C --exclude-nodes-with-label=autoscaler=disabled

D --filter-nodes-with-label=autoscaler=disabled

💡 Hint

Check official Cluster Autoscaler flags for ignoring nodes by label.

The correct flag is --skip-nodes-with-label to tell Cluster Autoscaler to ignore nodes with the given label.

❓ Troubleshoot

advanced

Why does Cluster Autoscaler not remove underutilized nodes?

Cluster Autoscaler is running but does not remove nodes even though some nodes have very low CPU and memory usage. What could be a reason?

A Cluster Autoscaler only adds nodes, it never removes nodes.

B There are pods with <code>podDisruptionBudget</code> preventing eviction on those nodes.

C Nodes with low usage are automatically removed by Kubernetes, no need for Cluster Autoscaler.

D Cluster Autoscaler requires manual approval to remove nodes.

💡 Hint

Think about what prevents pods from being moved or deleted.

Pods with podDisruptionBudget can block eviction, so Cluster Autoscaler cannot safely remove nodes hosting those pods.

🔀 Workflow

expert

Order the steps to enable Cluster Autoscaler on a new Kubernetes cluster

Put the steps in the correct order to enable Cluster Autoscaler on a new Kubernetes cluster.

A 1,2,3,4

B 3,1,2,4

C 2,1,3,4

D 1,3,2,4

💡 Hint

Think about what must exist before deploying Cluster Autoscaler and configuring it.

First create the node group, then configure flags, deploy Cluster Autoscaler, and finally verify logs.