KCNA Exam Prep Free practice test →

Free KCNA Practice Questions

10 free, exam-style Kubernetes and Cloud Native Associate (KCNA) practice questions with answers and explanations. No signup required. Work through them below, then take the full free KCNA practice test to study every exam domain.

Question 1

Which Kubernetes control plane component watches for newly created Pods that have not yet been assigned to a node, and then selects a suitable node for them to run on?

  1. kube-controller-manager
  2. kube-scheduler
  3. kubelet
  4. kube-apiserver
Show answer & explanation

Correct answer: B - kube-scheduler

Question 2

A platform team needs to run a logging agent on every node in the cluster, and the agent must automatically start on any new nodes that are added to the cluster later. Which Kubernetes object is designed for this requirement?

  1. Deployment
  2. StatefulSet
  3. DaemonSet
  4. ReplicaSet
Show answer & explanation

Correct answer: C - DaemonSet

Question 3

A Pod has a single container configured with a CPU request of 250m, a CPU limit of 500m, a memory request of 256Mi, and a memory limit of 256Mi. Which Quality of Service (QoS) class will Kubernetes assign to this Pod, and why?

  1. Burstable, because the CPU request and the CPU limit are not equal to each other
  2. Guaranteed, since the memory request equals the memory limit
  3. Guaranteed, since both requests and limits are set
  4. BestEffort, since the resources are not fully reserved
Show answer & explanation

Correct answer: A - Burstable, because the CPU request and the CPU limit are not equal to each other

Question 4

After Kubernetes removed the dockershim component (in version 1.24), which statement BEST describes how the kubelet runs containers on a worker node today?

  1. The kubelet communicates directly with the Docker daemon installed on every worker node
  2. The kubelet uses the Container Runtime Interface (CRI) to talk to containerd or CRI-O
  3. Container images that were originally built with Docker can no longer run on Kubernetes
  4. Kubernetes now runs each workload inside a lightweight virtual machine rather than a container
Show answer & explanation

Correct answer: B - The kubelet uses the Container Runtime Interface (CRI) to talk to containerd or CRI-O

Question 5

A newly scheduled Pod remains stuck in the ImagePullBackOff state. Which of the following is the MOST likely cause?

  1. The container starts and then immediately exits with an error, so the kubelet keeps restarting it
  2. The container requested more memory than its limit allowed and was terminated by the node
  3. No node in the cluster has enough free CPU or memory available to schedule the Pod
  4. The image cannot be retrieved because its name or tag is wrong, or the registry needs credentials
Show answer & explanation

Correct answer: D - The image cannot be retrieved because its name or tag is wrong, or the registry needs credentials

Question 6

In a namespace where no NetworkPolicy has ever been created, how can the Pods in that namespace communicate with one another by default?

  1. All Pods can freely send and receive traffic to and from any other Pod
  2. All Pods are isolated and cannot communicate until a NetworkPolicy explicitly permits it
  3. Pods can only communicate with other Pods that share the same labels
  4. Pods may receive incoming connections but cannot open outbound ones
Show answer & explanation

Correct answer: A - All Pods can freely send and receive traffic to and from any other Pod

Question 7

A team needs a single PersistentVolume to be mounted as read-write by Pods running on several different nodes at the same time. Which access mode must the PersistentVolume support?

  1. ReadWriteOnce (RWO)
  2. ReadOnlyMany (ROX)
  3. ReadWriteOncePod (RWOP)
  4. ReadWriteMany (RWX)
Show answer & explanation

Correct answer: D - ReadWriteMany (RWX)

Question 8

A cluster is managed using GitOps with Argo CD. An engineer manually edits a live Deployment directly in the cluster, bypassing the Git repository. Based on GitOps principles, what should happen?

  1. The manual change is automatically committed back into the Git repository
  2. The change is kept, because the live cluster state takes priority over Git
  3. The controller reconciles the cluster back to the state declared in Git
  4. Reconciliation is automatically paused until a human administrator reviews and approves the edit
Show answer & explanation

Correct answer: C - The controller reconciles the cluster back to the state declared in Git

Question 9

A team wants to release a new version of a service to a small percentage of real users first, observe how it behaves, and then gradually increase its share of traffic until it serves everyone. Which deployment strategy does this describe?

  1. Canary
  2. Blue-green
  3. Rolling update
  4. Recreate
Show answer & explanation

Correct answer: A - Canary

Question 10

By default, how does Prometheus - the CNCF graduated monitoring project - collect metrics from the applications it monitors?

  1. Each application continuously pushes its own metrics to Prometheus over a streaming connection
  2. Prometheus reads the metrics directly from each application's log files
  3. Prometheus periodically scrapes (pulls) metrics from HTTP endpoints exposed by targets
  4. Prometheus asks the Kubernetes API server to return each application's metrics
Show answer & explanation

Correct answer: C - Prometheus periodically scrapes (pulls) metrics from HTTP endpoints exposed by targets

Ready for the real thing?

Practice hundreds more KCNA questions with instant scoring, weak-area drills, and full exam simulations.

Start the free practice test See pricing