Goal of this exercise document

This exercise document complements the Intermediate workshop with practical tasks, solutions, and walkthroughs. The focus is on workload selection, Init Containers, sidecars, Ephemeral Containers, probes, graceful shutdown, PodDisruptionBudgets, requests, limits, QoS, LimitRange, ResourceQuota, scheduling, NetworkPolicy, DNS, Ingress or Gateway API, storage, Helm, Kustomize, and systematic troubleshooting.

Prerequisites

You need:

a lab cluster with a working CNI
CoreDNS
kubectl
optional: helm
optional: kustomize or kubectl kustomize
optional: a StorageClass
optional: an Ingress Controller or Gateway Controller

Recommended Namespace:

export NS=workshop-intermediate
kubectl create namespace "$NS"
kubectl config set-context --current --namespace="$NS"

Exercise overview

No.	Topic	Result
1	Workload selection	Deployment, Job, and CronJob distinguished technically
2	Init Container and sidecar	Pod lifecycles traced
3	Probes and lifecycle	Startup, Readiness, and Liveness cleanly separated
4	Resources, QoS, LimitRange, Quota	Namespace boundaries and scheduling impact checked
5	Scheduling	Node Labels, nodeSelector, Taints, and Tolerations tested
6	NetworkPolicy and DNS	Default Deny plus targeted allows created
7	Ingress or Gateway	HTTP app exposed externally or cleanly classified
8	PVC and StatefulSet	persistent state tested
9	Helm and Kustomize	rendered manifests compared
10	Troubleshooting course	typical errors systematically fixed

Part A – Exercises

Exercise 1: Select the workload resource

Goal

You decide whether Deployment, Job, or CronJob is technically appropriate.

Scenario

You need to model three requirements:

A web server must run continuously and be updated with rolling updates.
A data migration must run once and then finish.
A cleanup task must run every hour.

Task

Create a minimal manifest for each case.
Apply the manifests.
Check status and Events.
Document why the selected resource fits.

Exercise 2: Classify Init Container and sidecar

Goal

You understand when an Init Container must run before the application and when an additional container shares the same Pod lifecycle.

Task

Create a Pod where:

an Init Container writes a file to an emptyDir,
the App Container reads this file,
a sidecar container periodically writes a heartbeat file.

Check:

container startup order,
logs of all containers,
contents of the shared volume.

Exercise 3: Separate probes correctly

Goal

You correct a faulty probe configuration that causes restarts.

Scenario

A sample app starts slowly. The Liveness Probe checks the same path as Readiness too early.

Task

Create a Deployment with an intentionally too aggressive Liveness Probe.
Observe Events and restarts.
Add a Startup Probe.
Separate Readiness and Liveness conceptually.
Explain why Readiness controls traffic and Liveness triggers restarts.

Exercise 4: Resources, QoS, LimitRange, and ResourceQuota

Goal

You set requests and limits, observe QoS classes, and create Namespace boundaries.

Task

Create a LimitRange with default requests and limits.
Create a ResourceQuota for CPU, memory, and Pod count.
Roll out three Pods:
- without explicit resources,
- with requests lower than limits,
- with requests equal to limits.
Determine the QoS classes.
Intentionally create a Quota violation.

Exercise 5: Scheduling with Node Labels, nodeSelector, Taint, and Toleration

Goal

You control Pod placement and interpret Pending Events.

Task

Label a lab Node with workshop-role=apps.
Create a Pod with a matching nodeSelector.
Create a Pod with a non-matching nodeSelector.
Set a Taint on the Node.
Create one Pod without a Toleration and one with a Toleration.
Document the Events.

Exercise 6: NetworkPolicy Default Deny plus DNS allow

Goal

You create a Default Deny policy and then explicitly allow DNS and app traffic.

Task

Create two Deployments: frontend and backend.
Create Services for both.
Check that frontend can reach backend.
Create a Default Deny policy for Ingress and Egress.
Check that traffic is blocked.
Allow DNS Egress.
Allow frontend to backend on port 8080.
Check again.

Exercise 7: Provide Ingress or Gateway

Goal

You expose an HTTP application through Ingress or alternatively describe the Gateway API resources if no controller is installed.

Task

Check whether an Ingress Controller or Gateway Controller is present.
Deploy an HTTP app.
Create either an Ingress resource or a Gateway/HTTPRoute example.
Check controller Events, backend Service, and routing.
Document why an Ingress resource has no effect without a controller.

Exercise 8: Use PVC and StatefulSet

Goal

You run a small stateful workload with a PVC and check whether data survives Pod recreation.

Task

Check existing StorageClasses.
Create a StatefulSet with volumeClaimTemplates.
Write a file to the volume.
Delete the Pod.
After recreation, check whether the file still exists.
Document why a hostPath lab is not a production assumption.

Exercise 9: Compare Helm and Kustomize

Goal

You render manifests before applying them and compare Helm with Kustomize.

Task

Create a small Kustomize base and a dev overlay.
Render the overlay.
Create a minimal Helm chart or use a local sample chart.
Render the Helm chart.
Compare reviewability, parameterization, and error risk.

Exercise 10: Troubleshooting course

Goal

You examine five typical error patterns methodically.

Error patterns

Pod stays Pending because of an incorrect nodeSelector.
Pod shows ImagePullBackOff because of a wrong image tag.
Pod shows CrashLoopBackOff because of a faulty startup command.
PVC stays Pending because of an incorrect StorageClass.
Service has no Endpoints because of a wrong Selector.

Task

For each error, document:

symptom,
diagnostic command,
relevant Events or logs,
cause,
fix,
validation.

Part B – Solutions and walkthroughs

Solution 1: Select the workload resource

Deployment for a continuously running web server

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web
  namespace: workshop-intermediate
spec:
  replicas: 2
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
        - name: web
          image: nginx:1.27
          ports:
            - containerPort: 80

Job for a one-time migration

apiVersion: batch/v1
kind: Job
metadata:
  name: migration
  namespace: workshop-intermediate
spec:
  template:
    spec:
      restartPolicy: OnFailure
      containers:
        - name: migration
          image: busybox:1.36
          command: ["sh", "-c", "echo Migration running; sleep 3; echo Migration complete"]

CronJob for hourly cleanup

apiVersion: batch/v1
kind: CronJob
metadata:
  name: cleanup
  namespace: workshop-intermediate
spec:
  schedule: "0 * * * *"
  concurrencyPolicy: Forbid
  successfulJobsHistoryLimit: 3
  failedJobsHistoryLimit: 1
  jobTemplate:
    spec:
      template:
        spec:
          restartPolicy: OnFailure
          containers:
            - name: cleanup
              image: busybox:1.36
              command: ["sh", "-c", "date; echo cleanup"]

Check:

kubectl apply -f deployment.yml
kubectl apply -f job.yml
kubectl apply -f cronjob.yml
kubectl get deployment,job,cronjob,pod
kubectl logs job/migration

Technical decision:

Requirement	Resource	Reason
continuously running HTTP service	Deployment	replicas, rollouts, rollbacks
one-time migration	Job	completed state is technically desired
hourly cleanup	CronJob	time-based Job creation

Solution 2: Init Container and sidecar

Manifest:

apiVersion: v1
kind: Pod
metadata:
  name: lifecycle-demo
  namespace: workshop-intermediate
  labels:
    app: lifecycle-demo
spec:
  volumes:
    - name: shared
      emptyDir: {}
  initContainers:
    - name: init-write
      image: busybox:1.36
      command: ["sh", "-c", "echo prepared > /shared/init.txt"]
      volumeMounts:
        - name: shared
          mountPath: /shared
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "while true; do cat /shared/init.txt; sleep 30; done"]
      volumeMounts:
        - name: shared
          mountPath: /shared
    - name: sidecar
      image: busybox:1.36
      command: ["sh", "-c", "while true; do date > /shared/heartbeat.txt; sleep 10; done"]
      volumeMounts:
        - name: shared
          mountPath: /shared

Check:

kubectl apply -f lifecycle-demo.yml
kubectl get pod lifecycle-demo
kubectl describe pod lifecycle-demo
kubectl logs lifecycle-demo -c init-write
kubectl logs lifecycle-demo -c app
kubectl exec lifecycle-demo -c app -- ls -l /shared
kubectl exec lifecycle-demo -c app -- cat /shared/heartbeat.txt

Explanation:

Init Containers run before App Containers and must finish successfully.
Sidecars run in parallel with the application.
emptyDir shares files within the same Pod, but it does not survive Pod deletion.

Solution 3: Correct probes

Robust sample manifest:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: probe-demo
  namespace: workshop-intermediate
spec:
  replicas: 1
  selector:
    matchLabels:
      app: probe-demo
  template:
    metadata:
      labels:
        app: probe-demo
    spec:
      containers:
        - name: app
          image: registry.k8s.io/e2e-test-images/agnhost:2.53
          args: ["netexec", "--http-port=8080"]
          ports:
            - containerPort: 8080
          startupProbe:
            httpGet:
              path: /healthz
              port: 8080
            failureThreshold: 30
            periodSeconds: 2
          readinessProbe:
            httpGet:
              path: /healthz
              port: 8080
            initialDelaySeconds: 2
            periodSeconds: 5
          livenessProbe:
            httpGet:
              path: /healthz
              port: 8080
            initialDelaySeconds: 10
            periodSeconds: 10

Check:

kubectl apply -f probe-demo.yml
kubectl rollout status deployment/probe-demo
kubectl describe pod -l app=probe-demo
kubectl get pods -w

Sample explanation:

Startup Probe protects slow starts; while it is not successful, Liveness and Readiness are not active.
Readiness decides whether a Pod receives Endpoints for Services.
Liveness decides whether kubelet restarts the container.
A too aggressive Liveness Probe creates restart loops.

Solution 4: Resources, QoS, LimitRange, and ResourceQuota

LimitRange:

apiVersion: v1
kind: LimitRange
metadata:
  name: defaults
  namespace: workshop-intermediate
spec:
  limits:
    - type: Container
      defaultRequest:
        cpu: 100m
        memory: 128Mi
      default:
        cpu: 500m
        memory: 256Mi
      max:
        cpu: "1"
        memory: 512Mi

ResourceQuota:

apiVersion: v1
kind: ResourceQuota
metadata:
  name: namespace-quota
  namespace: workshop-intermediate
spec:
  hard:
    pods: "5"
    requests.cpu: "1"
    requests.memory: 1Gi
    limits.cpu: "3"
    limits.memory: 2Gi

Check:

kubectl apply -f limitrange.yml
kubectl apply -f quota.yml
kubectl describe limitrange defaults
kubectl describe resourcequota namespace-quota

Pod with Burstable QoS:

apiVersion: v1
kind: Pod
metadata:
  name: burstable-pod
  namespace: workshop-intermediate
spec:
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]
      resources:
        requests:
          cpu: 100m
          memory: 128Mi
        limits:
          cpu: 500m
          memory: 256Mi

Pod with Guaranteed QoS:

apiVersion: v1
kind: Pod
metadata:
  name: guaranteed-pod
  namespace: workshop-intermediate
spec:
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]
      resources:
        requests:
          cpu: 100m
          memory: 128Mi
        limits:
          cpu: 100m
          memory: 128Mi

Check QoS:

kubectl apply -f burstable-pod.yml
kubectl apply -f guaranteed-pod.yml
kubectl get pod burstable-pod -o jsonpath='{.status.qosClass}{"\n"}'
kubectl get pod guaranteed-pod -o jsonpath='{.status.qosClass}{"\n"}'

Create a Quota violation:

apiVersion: v1
kind: Pod
metadata:
  name: too-large
  namespace: workshop-intermediate
spec:
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]
      resources:
        requests:
          cpu: "2"
          memory: 2Gi
        limits:
          cpu: "2"
          memory: 2Gi

Expected result:

kubectl apply -f too-large.yml

The API Server rejects the resource because of ResourceQuota.

Solution 5: Scheduling

Select Node:

NODE=$(kubectl get nodes -o jsonpath='{.items[0].metadata.name}')
kubectl label node "$NODE" workshop-role=apps
kubectl get nodes --show-labels | grep workshop-role

Pod with matching nodeSelector:

apiVersion: v1
kind: Pod
metadata:
  name: scheduled-ok
  namespace: workshop-intermediate
spec:
  nodeSelector:
    workshop-role: apps
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]

Pod with non-matching nodeSelector:

apiVersion: v1
kind: Pod
metadata:
  name: scheduled-pending
  namespace: workshop-intermediate
spec:
  nodeSelector:
    workshop-role: does-not-exist
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]

Check:

kubectl apply -f scheduled-ok.yml
kubectl apply -f scheduled-pending.yml
kubectl get pods -o wide
kubectl describe pod scheduled-pending

Set Taint:

kubectl taint node "$NODE" dedicated=apps:NoSchedule

Pod with Toleration:

apiVersion: v1
kind: Pod
metadata:
  name: tolerated
  namespace: workshop-intermediate
spec:
  nodeSelector:
    workshop-role: apps
  tolerations:
    - key: dedicated
      operator: Equal
      value: apps
      effect: NoSchedule
  containers:
    - name: app
      image: busybox:1.36
      command: ["sh", "-c", "sleep 3600"]

Cleanup:

kubectl delete pod scheduled-ok scheduled-pending tolerated --ignore-not-found
kubectl taint node "$NODE" dedicated=apps:NoSchedule-
kubectl label node "$NODE" workshop-role-

Explanation:

nodeSelector is hard: no matching Node means Pending.
A Taint rejects Pods.
A Toleration allows a Pod to tolerate a Taint; by itself, it does not force placement.

Solution 6: NetworkPolicy and DNS

Sample app:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: backend
  namespace: workshop-intermediate
spec:
  replicas: 1
  selector:
    matchLabels:
      app: backend
  template:
    metadata:
      labels:
        app: backend
    spec:
      containers:
        - name: backend
          image: registry.k8s.io/e2e-test-images/agnhost:2.53
          args: ["netexec", "--http-port=8080"]
          ports:
            - containerPort: 8080
---
apiVersion: v1
kind: Service
metadata:
  name: backend
  namespace: workshop-intermediate
spec:
  selector:
    app: backend
  ports:
    - name: http
      port: 8080
      targetPort: 8080
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: frontend
  namespace: workshop-intermediate
spec:
  replicas: 1
  selector:
    matchLabels:
      app: frontend
  template:
    metadata:
      labels:
        app: frontend
    spec:
      containers:
        - name: frontend
          image: curlimages/curl:8.8.0
          command: ["sh", "-c", "sleep 3600"]

Test before Policy:

kubectl apply -f net-demo.yml
kubectl wait --for=condition=Available deployment/backend --timeout=90s
kubectl wait --for=condition=Available deployment/frontend --timeout=90s
FRONTEND=$(kubectl get pod -l app=frontend -o jsonpath='{.items[0].metadata.name}')
kubectl exec "$FRONTEND" -- curl -s http://backend:8080/hostname

Default Deny:

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: default-deny
  namespace: workshop-intermediate
spec:
  podSelector: {}
  policyTypes:
    - Ingress
    - Egress

Allow DNS Egress:

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: allow-dns-egress
  namespace: workshop-intermediate
spec:
  podSelector: {}
  policyTypes:
    - Egress
  egress:
    - to:
        - namespaceSelector:
            matchLabels:
              kubernetes.io/metadata.name: kube-system
      ports:
        - protocol: UDP
          port: 53
        - protocol: TCP
          port: 53

Allow frontend to backend:

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: allow-frontend-to-backend
  namespace: workshop-intermediate
spec:
  podSelector:
    matchLabels:
      app: backend
  policyTypes:
    - Ingress
  ingress:
    - from:
        - podSelector:
            matchLabels:
              app: frontend
      ports:
        - protocol: TCP
          port: 8080
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: allow-frontend-egress-to-backend
  namespace: workshop-intermediate
spec:
  podSelector:
    matchLabels:
      app: frontend
  policyTypes:
    - Egress
  egress:
    - to:
        - podSelector:
            matchLabels:
              app: backend
      ports:
        - protocol: TCP
          port: 8080

Check:

kubectl apply -f default-deny.yml
kubectl apply -f allow-dns-egress.yml
kubectl apply -f allow-frontend-backend.yml
kubectl exec "$FRONTEND" -- nslookup backend
kubectl exec "$FRONTEND" -- curl -s --max-time 5 http://backend:8080/hostname

Note:

NetworkPolicies are effective only if the CNI plugin enforces them.

Solution 7: Ingress or Gateway

Check controller:

kubectl get ingressclass
kubectl get gatewayclass 2>/dev/null || true
kubectl get pods -A | grep -Ei 'ingress|gateway|traefik|nginx|envoy|haproxy'

Ingress example:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: web
  namespace: workshop-intermediate
spec:
  rules:
    - host: web.localtest.me
      http:
        paths:
          - path: /
            pathType: Prefix
            backend:
              service:
                name: backend
                port:
                  number: 8080

Check:

kubectl apply -f ingress.yml
kubectl describe ingress web
kubectl get ingress web

Gateway API example if Gateway API is installed:

apiVersion: gateway.networking.k8s.io/v1
kind: HTTPRoute
metadata:
  name: web
  namespace: workshop-intermediate
spec:
  parentRefs:
    - name: shared-gateway
      namespace: gateway-system
  hostnames:
    - web.localtest.me
  rules:
    - backendRefs:
        - name: backend
          port: 8080

Technical explanation:

Ingress is only a rule description.
An Ingress Controller must translate that rule into proxy/load-balancer configuration.
Gateway API separates infrastructure responsibility more clearly from application routing.

Solution 8: PVC and StatefulSet

Check StorageClass:

kubectl get storageclass

StatefulSet:

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: data-demo
  namespace: workshop-intermediate
spec:
  serviceName: data-demo
  replicas: 1
  selector:
    matchLabels:
      app: data-demo
  template:
    metadata:
      labels:
        app: data-demo
    spec:
      containers:
        - name: app
          image: busybox:1.36
          command: ["sh", "-c", "while true; do sleep 3600; done"]
          volumeMounts:
            - name: data
              mountPath: /data
  volumeClaimTemplates:
    - metadata:
        name: data
      spec:
        accessModes: ["ReadWriteOnce"]
        resources:
          requests:
            storage: 1Gi

Apply and test:

kubectl apply -f statefulset.yml
kubectl rollout status statefulset/data-demo
kubectl exec data-demo-0 -- sh -c 'echo persistence-test > /data/test.txt'
kubectl delete pod data-demo-0
kubectl wait --for=condition=Ready pod/data-demo-0 --timeout=120s
kubectl exec data-demo-0 -- cat /data/test.txt

Expected state:

The file still exists after Pod recreation because the PVC is reattached to data-demo-0.

Classification:

StatefulSets provide stable Pod names and stable PVC assignment.
A local hostPath volume in a lab is convenient, but it is not portable and not production storage.

Solution 9: Helm and Kustomize

Kustomize structure:

kustomize-demo/
  base/
    deployment.yml
    service.yml
    kustomization.yml
  overlays/
    dev/
      kustomization.yml

base/kustomization.yml:

resources:
  - deployment.yml
  - service.yml

overlays/dev/kustomization.yml:

resources:
  - ../../base
namePrefix: dev-
labels:
  - pairs:
      environment: dev
images:
  - name: nginx
    newTag: "1.27"

Render:

kubectl kustomize kustomize-demo/overlays/dev
kubectl apply -k kustomize-demo/overlays/dev --dry-run=server

Minimal Helm chart:

helm create webchart
helm template web ./webchart
helm lint ./webchart

Comparison:

Criterion	Helm	Kustomize
Model	package with templates and values	base/overlay without template language
Strength	many parameters, releases, chart ecosystem	close to YAML, easy to review
Risk	complex templates can be hard to verify	overlays can become confusing
Mandatory check	`helm template`, `helm lint`	`kubectl kustomize`, `dry-run`

Solution 10: Troubleshooting course

Sample diagnosis:

Error pattern	Diagnosis	Cause	Fix
`Pending`	`kubectl describe pod`	no Node matches `nodeSelector`	correct Label or remove Selector
`ImagePullBackOff`	Events in the Pod	image/tag/registry wrong	set valid image
`CrashLoopBackOff`	`kubectl logs --previous`	process exits with an error	correct startup command/app configuration
`PVC Pending`	`kubectl describe pvc`	StorageClass missing or provisioner missing	correct or provide StorageClass
Service without Endpoints	`kubectl get endpoints`, check Labels	Selector does not match	correct Selector or Pod Labels

Standard sequence:

kubectl get all
kubectl get pod -o wide
kubectl describe pod <pod>
kubectl get events --sort-by=.metadata.creationTimestamp
kubectl logs <pod>
kubectl logs <pod> --previous
kubectl get svc,endpoints,endpointslices
kubectl describe pvc <pvc>

Cleanup

kubectl delete namespace workshop-intermediate

Assessment checklist

Criterion	Completed?
Workload resources distinguished technically	☐
Init Container, sidecar, and App Container explained	☐
Probes separated correctly	☐
Requests, limits, QoS, LimitRange, and Quota applied	☐
Scheduling Events interpreted	☐
NetworkPolicy including DNS considered	☐
Ingress/Gateway Controller role explained	☐
PVC and StatefulSet tested	☐
Helm and Kustomize rendered	☐
Troubleshooting course documented	☐

Kubernetes Workshop: Intermediate – Exercises with Solutions

Goal of this exercise document

Prerequisites

Exercise overview

Part A – Exercises

Exercise 1: Select the workload resource

Goal

Scenario

Task

Exercise 2: Classify Init Container and sidecar

Goal

Task

Exercise 3: Separate probes correctly

Goal

Scenario

Task

Exercise 4: Resources, QoS, LimitRange, and ResourceQuota

Goal

Task

Exercise 5: Scheduling with Node Labels, nodeSelector, Taint, and Toleration

Goal

Task

Exercise 6: NetworkPolicy Default Deny plus DNS allow

Goal

Task

Exercise 7: Provide Ingress or Gateway

Goal

Task

Exercise 8: Use PVC and StatefulSet

Goal

Task

Exercise 9: Compare Helm and Kustomize

Goal

Task

Exercise 10: Troubleshooting course

Goal

Error patterns

Task

Part B – Solutions and walkthroughs

Solution 1: Select the workload resource

Deployment for a continuously running web server

Job for a one-time migration

CronJob for hourly cleanup

Solution 2: Init Container and sidecar

Solution 3: Correct probes

Solution 4: Resources, QoS, LimitRange, and ResourceQuota

Solution 5: Scheduling

Solution 6: NetworkPolicy and DNS

Solution 7: Ingress or Gateway

Solution 8: PVC and StatefulSet

Solution 9: Helm and Kustomize

Solution 10: Troubleshooting course

Cleanup

Assessment checklist

31 modules ως ενιαία διαδρομή μάθησης.