Certified Kubernetes Application Developer

The blueprint

A mock lab here has 16 tasks in 2 hours, split across the domains in the same proportions as the official exam guide. The official pass mark is 66%. FixOps scores practice against a target of 66%.

1. 01

   Application Design and Build

   20%

2. 02

   Application Deployment

   20%

3. 03

   Application Observability and Maintenance

   15%

4. 04

   Application Environment, Configuration and Security

   25%

5. 05

   Services and Networking

   20%

The free sample lab

Two tasks from the bank, in a simulated cluster you drive with kubectl and helm. You are graded on the state you leave behind, not on the commands you type.

• Add a log-shipping sidecar. Create a two-container pod whose containers share a volume.
• A crash loop with the answer in the logs. Read the logs of a crash-looping Deployment and supply the setting it asks for.

Revision notes: Application Design and Build

The notes for one domain, free to read here and in the app. FixOps Pro has them for all 5 domains.

Building images, choosing the right workload resource, composing multi-container pods and using volumes.

Images

• A Dockerfile's FROM names the base image, COPY adds files, RUN executes at build time, and CMD or ENTRYPOINT is what runs at start.
• Multi-stage builds compile in one stage and copy only the result into a small final image.
• In a pod, command replaces the image's ENTRYPOINT and args replaces its CMD.
• Pin images by a specific tag or digest; imagePullPolicy defaults to Always for the latest tag and IfNotPresent otherwise.

Choosing a workload

• Deployment for stateless replicas, StatefulSet for stable identity and per-pod storage, DaemonSet for one pod per node.
• kubectl create job NAME --image=IMAGE -- COMMAND runs to completion; completions and parallelism control how many pods and how many at once; backoffLimit caps retries.
• kubectl create cronjob NAME --image=IMAGE --schedule="*/15 * * * *" -- COMMAND runs Jobs on a schedule.
• A CronJob's concurrencyPolicy is Allow, Forbid or Replace, and the history limits keep a number of finished Jobs.
• A Job's pods need restartPolicy Never or OnFailure.

Multi-container pods

• Containers in a pod share the network (localhost) and any volume both of them mount, but not their file systems.
• A sidecar extends the main container, for example by shipping its logs from a shared emptyDir.
• Init containers run one after another to completion before the application containers start.
• An init container with restartPolicy: Always is a native sidecar: it starts first and keeps running beside the main containers.
• kubectl logs POD -c CONTAINER and kubectl exec POD -c CONTAINER -- COMMAND address one container.

Volumes

• emptyDir is scratch space for the life of the pod and the usual way to share files between containers.
• A PersistentVolumeClaim requests durable storage; the pod mounts the claim by name.
• volumeMounts in each container say where a volume appears; the same volume can have different paths in different containers.
• readOnly: true on a mount protects the data from that container.

Easy to mix up

• command overrides ENTRYPOINT; args overrides CMD.
• Init containers finish before the app starts; sidecars run alongside it.
• Forbid skips a new CronJob run while the last is still going; Replace kills the old one.
• completions is how many pods must succeed; parallelism is how many run at the same time.