KCNA — Kubernetes Architecture Practice Questions
Kubernetes Architecture Practice Questions
Test your understanding of Kubernetes architecture — control plane, worker nodes, API, networking, and container runtimes. This chapter covers the Kubernetes Fundamentals domain (46% of the exam).
Question 1
Which control plane component is the only one that directly reads and writes to etcd?
- A) kube-scheduler
- B) kube-controller-manager
- C) kube-apiserver
- D) kubelet
Show Answer
Correct Answer: C
| Option | Explanation |
|---|---|
| A | kube-scheduler communicates with the API server, not etcd directly |
| B | kube-controller-manager communicates with the API server, not etcd directly |
| C | kube-apiserver is the ONLY component that reads from and writes to etcd directly. All other components go through the API server. |
| D | kubelet runs on worker nodes and communicates with the API server |
Question 2
What happens when the kube-scheduler assigns a pod to a node?
- A) The scheduler starts the container on the node
- B) The scheduler notifies the kubelet on that node via the API server
- C) The scheduler pulls the container image to the node
- D) The scheduler creates the pod directly on the node
Show Answer
Correct Answer: B
| Option | Explanation |
|---|---|
| A | The scheduler does NOT start containers — kubelet does that |
| B | The scheduler updates the pod’s node assignment in etcd (via API server). kubelet on that node watches for assignments and starts the container. |
| C | Image pulling is done by the container runtime (containerd), not the scheduler |
| D | The scheduler never directly interacts with nodes — everything goes through the API server |
Question 3
Which component maintains network rules for routing Service traffic to pods?
- A) CoreDNS
- B) kubelet
- C) kube-proxy
- D) CNI plugin
Show Answer
Correct Answer: C
| Option | Explanation |
|---|---|
| A | CoreDNS resolves service names to ClusterIP addresses — it doesn’t route traffic |
| B | kubelet manages pod lifecycle, not network routing |
| C | kube-proxy maintains iptables or IPVS rules on each node to route Service (ClusterIP) traffic to the correct backend pods |
| D | CNI plugins handle pod-to-pod networking (IP assignment, overlay), not Service routing |
Question 4
What is the default container runtime in most Kubernetes distributions since v1.24?
- A) Docker
- B) CRI-O
- C) containerd
- D) runc
Show Answer
Correct Answer: C
| Option | Explanation |
|---|---|
| A | Docker (dockershim) was removed in Kubernetes v1.24 |
| B | CRI-O is a valid runtime but is mainly used by OpenShift — not the default for most distributions |
| C | containerd is the default container runtime for most K8s distributions including EKS and GKE. It was always inside Docker and is now used directly. |
| D | runc is a low-level OCI runtime — it’s used by containerd/CRI-O internally, but it’s not a CRI runtime itself |
Question 5
In Kubernetes networking, how do containers within the same pod communicate?
- A) Through the CNI plugin
- B) Through kube-proxy
- C) Through localhost
- D) Through the Service mesh
Show Answer
Correct Answer: C
| Option | Explanation |
|---|---|
| A | CNI handles pod-to-pod networking across nodes, not container-to-container within a pod |
| B | kube-proxy handles Service-to-pod routing, not intra-pod communication |
| C | Containers in the same pod share the same network namespace, so they communicate via localhost (127.0.0.1) using different port numbers |
| D | Service mesh handles service-to-service communication, not container-to-container within a pod |
Question 6
What does CRI stand for in Kubernetes?
- A) Container Registry Interface
- B) Container Runtime Interface
- C) Container Resource Interface
- D) Cluster Runtime Interface
Show Answer
Correct Answer: B
| Option | Explanation |
|---|---|
| A | Container registries (Docker Hub, ECR) store images — CRI is not about registries |
| B | CRI (Container Runtime Interface) is the API standard that defines how kubelet communicates with container runtimes like containerd and CRI-O |
| C | Resource management is handled by kubelet and the scheduler, not by CRI |
| D | CRI operates at the node level, not the cluster level |
Question 7
If etcd completely fails in a Kubernetes cluster, what is the impact?
- A) Existing running pods stop immediately
- B) The cluster cannot schedule new pods or process any changes
- C) Worker nodes disconnect from the cluster
- D) All container runtimes crash
Show Answer
Correct Answer: B
| Option | Explanation |
|---|---|
| A | Existing running pods continue to run — kubelet manages them independently. But no new changes can be processed. |
| B | etcd stores all cluster state. Without it, the API server cannot read or write state, so no new pods can be scheduled, no updates can be processed, and no changes can be made. |
| C | Worker nodes and kubelets continue operating — they just can’t receive new instructions |
| D | Container runtimes are independent of etcd — they continue running existing containers |
Question 8
Which DNS server is used by default in Kubernetes clusters for service discovery?
- A) BIND
- B) dnsmasq
- C) CoreDNS
- D) Route 53
Show Answer
Correct Answer: C
| Option | Explanation |
|---|---|
| A | BIND is a general-purpose DNS server — not used in Kubernetes by default |
| B | dnsmasq was used in older Kubernetes versions but has been replaced |
| C | CoreDNS is the default DNS server in Kubernetes (CNCF Graduated). It resolves service names to ClusterIP addresses for service discovery. |
| D | Route 53 is an AWS DNS service — not a Kubernetes component |
Question 9
Which Kubernetes controller detects when a node goes down and marks it as NotReady?
- A) Replication Controller
- B) Node Controller
- C) Endpoint Controller
- D) Deployment Controller
Show Answer
Correct Answer: B
| Option | Explanation |
|---|---|
| A | Replication Controller ensures the correct number of pod replicas are running |
| B | The Node Controller monitors node health by checking heartbeats. When a node stops responding, it marks it as NotReady and eventually evicts pods. |
| C | Endpoint Controller manages the association between Services and pods |
| D | Deployment Controller manages the lifecycle of Deployments and their ReplicaSets |
Question 10
What are the three plugin interfaces (CRI, CNI, CSI) used for in Kubernetes?
- A) Compute, Networking, Storage
- B) Configuration, Naming, Security
- C) Containers, Nodes, Services
- D) Runtime, Networking, Storage
Show Answer
Correct Answer: D
| Option | Explanation |
|---|---|
| A | “Compute” is too vague — CRI specifically deals with container runtimes |
| B | These don’t map to CRI, CNI, or CSI |
| C | These don’t accurately describe the three interfaces |
| D | CRI = Container Runtime Interface, CNI = Container Network Interface, CSI = Container Storage Interface. They standardize runtime, networking, and storage plugins respectively. |
Score Yourself
| Score | Rating | Action |
|---|---|---|
| 9-10 | Excellent | Strong architecture knowledge. Move to Chapter 3. |
| 7-8 | Good | Review the topics you missed, then proceed. |
| 5-6 | Fair | Re-read control plane and worker node sections. |
| Below 5 | Needs Work | Go back through all Architecture sub-topics. |
What’s Next?
You’ve completed Chapter 2 — Kubernetes Architecture. Now let’s learn about the resources you create and manage in a cluster.
