Interview Prep Guide

Kubernetes Production Operations Interview Questions

Kubernetes questions on scheduling, resources, networking, autoscaling, security, rollouts, stateful workloads, and incidents.

Applied Kubernetes Questions

  1. How do requests and limits affect Kubernetes scheduling and runtime?

    Requests guide scheduling and reserved capacity; limits constrain runtime usage and can cause CPU throttling or memory termination.

  2. When should you use a Deployment, StatefulSet, DaemonSet, or Job?

    Choose by workload identity and lifecycle: interchangeable replicas, stable stateful identity, one Pod per node, or finite retryable work.

  3. How does a Kubernetes Service route traffic to changing Pods?

    A Service selects ready endpoints and provides stable discovery while platform networking routes traffic to eligible Pods.

  4. How should ConfigMaps and Secrets be updated safely?

    Treat configuration as versioned rollout input, restrict secret access, avoid accidental logs, and ensure workloads reload or restart predictably.

  5. What do PodDisruptionBudgets protect against?

    They limit simultaneous voluntary disruptions so enough replicas remain available during drains or controlled maintenance.

Advanced Kubernetes Questions

  1. How do horizontal autoscaling signals and delays affect reliability?

    Autoscaling follows measured signals with collection, stabilization, scheduling, and startup delays, so it cannot absorb every sudden spike instantly.

  2. How would you enforce least privilege for Kubernetes workloads?

    Use dedicated service accounts, narrow RBAC, restricted Pod security, network policies, secret boundaries, admission policy, and auditable exceptions.

  3. What makes a Kubernetes rollout safe?

    Use readiness and startup behavior, bounded surge and unavailability, progressive exposure, compatibility, observability, and an exercised rollback.

  4. How do affinity, taints, tolerations, and topology spread differ?

    Affinity expresses placement preference or requirement, taints repel Pods without tolerations, and topology spread balances replicas across failure domains.

  5. How should persistent workloads be designed in Kubernetes?

    Match StatefulSets, storage classes, backup, topology, recovery, and application consistency requirements instead of assuming volume attachment equals durability.

Kubernetes Production Scenarios

  1. Pods are Running but the service returns no traffic. What do you inspect?

    Check readiness, Service selectors, endpoints, ports, DNS, network policy, ingress, and application bind addresses in a layered order.

  2. A rollout causes intermittent 502 errors. How do you diagnose it?

    Correlate errors with Pod lifecycle, readiness timing, connection draining, termination grace, load balancer health, and version compatibility.

  3. A node is under memory pressure and Pods are being evicted. What do you do?

    Identify workload consumption, QoS and eviction order, incorrect requests or limits, node capacity, and whether a leak or traffic shift caused pressure.

  4. The cluster cannot schedule new Pods despite low average CPU. Why might that happen?

    Requests, per-node fit, memory, storage, ports, affinity, taints, topology rules, quotas, or fragmented capacity can block placement.

Additional Frequently Tested Questions

  1. How do PodDisruptionBudgets affect maintenance and availability?

    They limit how many selected Pods may be voluntarily disrupted at once, helping controllers preserve an intended level of availability.