Why Every Ops Role Now Demands Kubernetes Skills (And a 100‑Question K8s Interview Guide)

The author, after five years in operations and being laid off, discovered that virtually every job posting now expects proficiency in Docker and Kubernetes, prompting the recommendation to learn these technologies as soon as possible.

To help job seekers, the author shares a compiled interview document titled "100 K8s Interview Questions (High‑Salary Essential)" that systematically covers the key knowledge areas required for Kubernetes interviews.

1. Kubernetes Basics and Architecture

K8s is what: an open‑source container orchestration platform for automating deployment, scaling, and management of containerized applications.

Core components:

Fundamental objects: Pod, Service, Deployment, Namespace, Label, Volume, ConfigMap, Secret, etc.

2. Core Pod Knowledge

Pod is the smallest scheduling unit and can contain one or more containers.

Lifecycle and states: Pending, Running, Succeeded, Failed, Unknown.

Restart policies: Always, OnFailure, Never.

Image pull policies: Always, IfNotPresent, Never.

Health‑check mechanisms: LivenessProbe, ReadinessProbe, StartupProbe.

Pod creation flow: API request → scheduling → binding → container creation → status sync.

Pod termination flow: graceful shutdown, signal handling, resource cleanup.

3. Resource Management and Scheduling

Deployment & ReplicaSet: manage pod replicas and perform rolling updates.

DaemonSet: runs a pod on every node, useful for logging, monitoring, etc.

Job: one‑time task; pod does not restart after completion.

Resource requests & limits: define CPU/memory needs.

QoS classes: Guaranteed, Burstable, BestEffort.

Scheduling mechanisms: predicates, priorities, node affinity, taints & tolerations.

4. Networking and Service Discovery

Service types: ClusterIP, NodePort, LoadBalancer, ExternalName.

Headless Service: no ClusterIP, suitable for stateful services.

Discovery methods: environment variables, DNS.

kube‑proxy implementation: iptables or IPVS based load balancing.

Network plugins: Flannel (simple overlay), Calico (BGP routing), OVS (multi‑tenant support).

Ingress: HTTP/HTTPS layer routing with domain‑based forwarding.

5. Storage and Data Persistence

Volume types: emptyDir, hostPath, configMap, secret, downwardAPI, projected, PVC, etc.

PV & PVC: PersistentVolume is a cluster resource; PersistentVolumeClaim is a user’s storage request.

Access modes: RWO, ROX, RWX.

Reclaim policies: Retain, Delete, Recycle (deprecated).

CSI model: standardized storage plugin interface that decouples Kubernetes from storage providers.

6. Security and Access Control

Authentication & authorization: RBAC, ServiceAccount, Secret, Token.

Admission control: PodSecurityPolicy, LimitRanger, NamespaceExists, etc.

Pod security policies: restrict privileged mode, host access, user IDs, etc.

7. Operations and Troubleshooting

Common reasons for a pod staying Pending: scheduling failures (insufficient resources, affinity conflicts) or PVC creation failures.

kubectl drain: gracefully evicts pods and removes a node for maintenance.

etcd data sync: uses the Raft protocol to ensure consistency.

Production issue investigation steps:

8. Add‑on Components and Ecosystem Tools

Helm: package manager for Kubernetes, supports templated deployments.

EFK (Elasticsearch + Fluentd + Kibana): log collection and visualization.

Metrics Server: provides core metrics such as CPU and memory.

Cluster federation: unified management of multiple Kubernetes clusters.