Preparing this page…
Your current navigation and account state stay available while the next page loads.
Practice Docker and Kubernetes interview scenarios in a guided browser simulation. No containers, clusters, cloud accounts, or arbitrary workloads are created.
A Deployment declares desired replicas and Pod template. Its controller manages ReplicaSets and Pods until actual state converges.
A Service selector continuously matches Pods; matching Ready Pod addresses appear in EndpointSlices and can receive traffic.
Failed readiness removes a Pod from Service endpoints. Failed liveness can restart a container; the two signals solve different problems.
Changing a Deployment Pod template creates a rollout revision. Scaling alone does not; rollout status and undo operate on revisions.
Clients use the Kubernetes API through kube-apiserver. The API server persists serialized API-object state in etcd; ordinary workloads and kubectl do not bypass it to edit etcd directly.
NetworkPolicy selects Pods and allows specified layer 3 or 4 traffic. Enforcement depends on a cluster network plugin that supports NetworkPolicy.
Each Pod receives a cluster-wide IP. The pod network must let Pods communicate directly across nodes unless intentional segmentation such as NetworkPolicy restricts them.
A PVC requests storage and can be mounted into a Pod. Its bound PersistentVolume has a lifecycle independent of any one consuming Pod.
Follow these in order for the shortest path through the official beginner concepts.
Identify the active control plane, worker nodes, namespaces, and current kubectl context.
Start lessonCreate a Deployment, inspect its Pod, read logs, and clean up the workload.
Start lessonCreate non-confidential configuration, reference it from a Pod, and verify the injected value.
Start lessonObserve how selectors, Ready Pods, and endpoints determine request flow.
Start lessonChange desired replicas and observe the Deployment, Pods, and Service endpoints converge.
Start lessonThe command interpreter runs in a Web Worker separated from the page DOM. It accepts only bounded, structured lesson messages, and the page validates returned state before saving it locally. It does not run Docker, Kubernetes, shell commands, images, YAML, networks, or user code. Output is lesson-authored and deterministic.