Problems

Labs open a real Linux machine in your browser. Quizzes check the theory in a minute.

22 of 277 problems

Title	Topic	Acceptance	Difficulty	XP
Fix a Deployment Stuck Pulling Its ImageA rollout is stuck with Pods in ImagePullBackOff. Find the broken image reference and get every replica running.easyKubernetes10 XP	Kubernetes	New	easy	10
Fix a Service With No EndpointsA Service routes to nothing because its selector misses the Pods. Point it at exactly the right replicas without catching the canary.easyKubernetes10 XP	Kubernetes	New	easy	10
Fix Pods Stuck in PendingA Deployment asks for more CPU than any node has. Find out why its Pods never start and right-size the requests.easyKubernetes10 XP	Kubernetes	New	easy	10
Add Readiness and Liveness ProbesGive a Deployment the health checks it is missing, so Kubernetes stops sending traffic to Pods that are not ready and restarts hung ones.mediumKubernetes20 XP	Kubernetes	New	medium	20
Give a ServiceAccount Read-Only AccessCreate a ServiceAccount that can watch Pods and Deployments in one namespace and nothing more.mediumKubernetes20 XP	Kubernetes	New	medium	20
Build a Cluster-Wide Read-Only RoleGive an auditor read access to nodes, volumes and Pods across the cluster, without Secrets and without any write access.hardKubernetes40 XP	Kubernetes	New	hard	40
Pending on Purpose, Pending by MistakeA batch workload must run on the node reserved for it, but its Pods never get placed. Fix the workload, not the nodes.mediumKubernetes20 XP	Kubernetes	New	medium	20
A Deployment That Creates No PodsA Deployment sits at zero Pods and there are no Pending Pods to describe. Find what blocks it and fix the Deployment.mediumKubernetes20 XP	Kubernetes	New	medium	20
Stuck in PendingA new release never starts; its Pods sit in Pending. Read the scheduler's events, find the cause and fix it the right way.mediumKubernetes20 XP	Kubernetes	New	medium	20
Keep Pods Away From the Metadata ServicePods can reach the cloud's instance metadata endpoint and steal the node's credentials. Block it for every Pod in the namespace, without cutting off anything else.mediumKubernetes20 XP	Kubernetes	New	medium	20
Only Images From the Company RegistryWrite a ValidatingAdmissionPolicy that rejects Pods with images from anywhere but the company registry, in the namespaces that opt in.hardKubernetes40 XP	Kubernetes	New	hard	40
What Is a Pod?Which statement about a Kubernetes Pod is correct?easyKubernetes3 XP	Kubernetes	New	easy	3
Service TypesWhich Service type gives Pods a stable cluster-internal virtual IP and DNS name, and is the default?easyKubernetes3 XP	Kubernetes	New	easy	3
Pods That Come BackYou run kubectl delete pod metrics-node01 and it reappears within seconds with the same name. It has no ownerReferences and its name ends with the node name. HoeasyKubernetes3 XP	Kubernetes	New	easy	3
A Service with No Endpointskubectl get endpointslices -l kubernetes.io/service-name=web shows no endpoints, although Pods of the web app are Running and Ready. What is the most likely caumediumKubernetes5 XP	Kubernetes	New	medium	5
Role vs ClusterRoleA ServiceAccount in namespace ci must list Pods only in namespace staging. Which combination is correct?mediumKubernetes5 XP	Kubernetes	New	medium	5
CrashLoopBackOffA Pod is in CrashLoopBackOff. Which command most directly shows why the previous container instance exited?mediumKubernetes5 XP	Kubernetes	New	medium	5
NetworkPolicy DefaultsA namespace has no NetworkPolicies. You then add one policy that selects Pods app=db and allows ingress only from Pods app=api. Which statement is correct (withhardKubernetes8 XP	Kubernetes	New	hard	8
ResourceQuota Needs DefaultsAn admin added a ResourceQuota on requests.cpu and limits.cpu to namespace team-a. Now a Deployment there has 0 Pods and kubectl get pods shows nothing. kubectlhardKubernetes8 XP	Kubernetes	New	hard	8
Pinning Images by DigestCI scans registry.example.com/api:1.8 with Trivy and it passes. Later someone pushes a different image under the same tag. How do you make sure Pods run exactlyeasyKubernetes3 XP	Kubernetes	New	easy	3
HPA, VPA and Cluster AutoscalerDuring a traffic spike, the HorizontalPodAutoscaler scales api from 4 to 20 replicas, but 6 new Pods stay Pending with 0/5 nodes are available: Insufficient cpumediumKubernetes5 XP	Kubernetes	New	medium	5
Upgrades That Undo ThemselvesYou want an upgrade to roll back automatically if the new Pods never become ready within 5 minutes. Which command does that?mediumKubernetes5 XP	Kubernetes	New	medium	5