Skip to content

05_kubernetes.md

Latest commit

 

History

History
206 lines (149 loc) · 6.3 KB

05_kubernetes.md

File metadata and controls

206 lines (149 loc) · 6.3 KB
Raw

05 – Kubernetes

Core K8s topics: pods, services, ingress, CRDs, and debugging.

Table of Contents

  1. Deploying a New Application
  2. Services vs Ingress
  3. CRDs & Operators
  4. Logs & Crash Troubleshooting
  5. Resource in “Terminating” State
  6. Debugging Inside a Container
  7. Editing a Resource Live
  8. Service-to-Service Communication
  9. Sidecar/Init Containers
  10. Bonus: Resource Name Limits
  11. Scenario: RBAC Permissions for Kaniko Builds
  12. Scenario: Kubernetes Pod Logs Lost After Crash
  13. Scenario: Resource Exhaustion (OOMKills) in Pods

1) Deploying a New Application

Question:
If you have a new microservice to run in Kubernetes, what resources do you usually set up?

Hints / Key Points
  • Usually a Deployment (or StatefulSet if stateful) and a Service.
  • Possibly an Ingress or LoadBalancer if external access is required.
  • Might use Helm for templating.

2) Services vs Ingress

Question:
What does a Service do in Kubernetes, and how is it different from an Ingress?

Hints / Key Points
  • Service: Exposes pods at a stable address, can be ClusterIP, NodePort, or LoadBalancer.
  • Ingress: Defines routing rules for HTTP/HTTPS traffic to one or more Services.

3) CRDs & Operators

Question:
How do CRDs (Custom Resource Definitions) and Operators help you extend Kubernetes beyond its default features?

Hints / Key Points
  • A CRD adds a new type of object (like “MyDatabase”) to the cluster.
  • An Operator watches these CRDs and automates tasks (install, upgrade, manage).
  • Good for complex/stateful apps so K8s can handle them more natively.

4) Logs & Crash Troubleshooting

Question (Scenario):
Your app keeps crashing after a few minutes in Kubernetes. How would you check what’s going on?

Hints / Key Points
  • Inspect logs from the pod/container.
  • Check events or error messages for the pod.
  • See if it’s an OOM kill, code exception, or config problem.

5) Resource in “Terminating” State

Question:
Sometimes a pod or other resource is stuck “Terminating” for a long time. Why could that happen, and what might you do?

Hints / Key Points
  • Finalizers might be blocking deletion.
  • The app might not handle termination signals well, so it never exits.
  • You can remove the finalizer or do a force delete if absolutely needed.

6) Debugging Inside a Container

Question:
You need to run commands inside a container for debugging. How do you do that in a Kubernetes environment?

Hints / Key Points
  • Typically use a CLI to exec into the container.
  • If multiple containers, specify which container.
  • Make sure you have the right RBAC privileges.

7) Editing a Resource Live

Question (Scenario):
You spot a small config mistake in a live resource. How would you fix it right away in the cluster? What risks might that cause?

Hints / Key Points
  • You can edit the resource in place with the CLI, but that can cause drift from Git or Helm config.
  • If you’re using GitOps, the next sync might overwrite your manual fix.
  • Best practice: fix it in your config repo or chart too.

8) Service-to-Service Communication

Question:
How do different services within the same cluster talk to each other?

Hints / Key Points
  • Cluster DNS: <service-name>.<namespace>.svc.cluster.local.
  • A Service provides a stable endpoint, even if pod IPs change.

9) Sidecar/Init Containers

Question:
What are sidecar containers and init containers, and why might you use them?

Hints / Key Points
  • Init containers run first to do setup tasks (migrations, config).
  • Sidecar containers run alongside the main app for logging, proxying, etc.
  • Helps separate concerns in a single pod.

10) Bonus: Resource Name Limits

Question:
Is there a name length limit or other format rule for K8s resources?

Hints / Key Points
  • Usually follows DNS label rules (lowercase, up to 63 chars, alphanumeric + dashes).
  • Some resource types might vary slightly, but typically the same constraints apply.

11) Scenario: RBAC Permissions for Kaniko Builds

Question:
You’re running an Azure DevOps agent in Kubernetes, which uses Kaniko to build and push Docker images. It’s failing because it can’t create needed resources.

  • How would you troubleshoot the missing permissions?
  • How can RBAC be configured to give Kaniko the required access?
Hints / Key Points
  • Check the Pod logs for permission errors (e.g., “forbidden”).
  • Assign a ServiceAccount with an appropriate Role/RoleBinding that allows creating ConfigMaps, Pods, etc.
  • Verify that the agent is using this ServiceAccount when building.

12) Scenario: Kubernetes Pod Logs Lost After Crash

Question:
A Kubernetes Pod crashes unexpectedly, and its logs are lost because the container restarts too quickly.

  • How would you recover logs from a previously crashed container?
  • How can you ensure logs are always accessible?
Hints / Key Points
  • Use the CLI to get logs from the previous container instance (-p option), if still available.
  • Centralize logs in an external system like ELK, Loki, or FluentD.
  • Ensure your app flushes logs frequently so they aren’t lost on crash.

13) Scenario: Resource Exhaustion (OOMKills) in Pods

Question:
A Kubernetes Pod crashes intermittently and is marked as OOMKilled.

  • How would you identify the cause of the memory spikes?
  • How do you stop the Pod from running out of memory in the future?
Hints / Key Points
  • Check resource usage with kubectl top or a monitoring tool.
  • Increase the memory limit if the app truly needs more, or find memory leaks.
  • Monitor usage over time, maybe use VPA (Vertical Pod Autoscaler) if appropriate.