OpenLabs handbook

Kubernetes Visual Handbook

Declare a workload. Watch the system reconcile it.

Edit a native manifest and inspect the difference between desired state, Pod identity, container restarts and node reservations. Interactive state models, with no flow engine required.

Roadmap

The order in which the concepts hold each other up. The interactive map is not drawn yet — drawing the path before walking it would be a promise, so the phases are listed plainly instead.

01 A workload becomes real The manifest, the owner, the Pod and the node are different things.
02 Keeping the promise Updates, availability and the reconciliation decisions still to explore.

Articles and labs

Each phase pairs prose with a model. The article says what is true; the lab is the thing running, and you can break it.

01 A workload becomes real

The manifest, the owner, the Pod and the node are different things.

Articles

A Pod is not its container ready

Labs

A manifest, a Pod, and the controller that replaces it ready

02 Keeping the promise

Updates, availability and the reconciliation decisions still to explore.

Articles

A rollout is not a restart coming

Labs

What remains available during an update coming

HANDBOOK ASSESSMENT · BANK 0.1.0

Kubernetes handbook assessment

A cumulative practice assessment of the currently available Pod lifecycle material. Future handbook phases are not covered yet.

7 questions · 14 points · target 70%

Practice bank · educator review pending for 7 questions.

QUESTION 1 · single-choice1 point

A container is OOM-killed and restarted inside a Pod. What happens to the Pod UID?

Choose one answer.

A container is OOM-killed and restarted inside a Pod. What happens to the Pod UID?The Pod keeps its UID.Every container restart creates a new Pod UID.The existing Pod moves to another node.

QUESTION 2 · multiple-response2 points

Select every correct statement about requests and limits in this lab.

Select all that apply. Credit requires exactly the correct set.

Select every correct statement about requests and limits in this lab.CPU and memory requests are used for placement.Exceeding a memory limit can cause the container to be OOM-killed.Raising a CPU limit creates extra replicas.A CPU request is a measurement of current CPU usage.

QUESTION 3 · ordering2 points

Order a successful startup in this lab, from creation to ready. Its model uses two fixed startup steps and no probes.

Drag items into order, or use the Move up/down buttons. Each correct position earns a share of the points.

Pod becomes Ready in the lab
ReplicaSet creates the Pod object
Kubelet starts the container
Scheduler binds the Pod to a node

QUESTION 4 · matching3 points

Match each responsibility to the component that performs it. Targets may be used more than once.

Drag a responsibility to a target, or choose its target in the list. Targets can be reused. Each correct match earns a share of the points.

Creates a replacement to maintain the replica countSelects a node for a new unscheduled PodEnsures the container starts on the assigned node

Kubelet

ReplicaSet controller

Scheduler

QUESTION 5 · code-analysis3 points

Each of the two lab nodes has 2000m allocatable CPU and no reservations. What happens to a Pod using this container resource fragment?

Choose one answer.

resources:
  requests:
    cpu: 3000m
    memory: 256Mi
  limits:
    cpu: 4000m
    memory: 512Mi

Each of the two lab nodes has 2000m allocatable CPU and no reservations. What happens to a Pod using this container resource fragment?It remains Pending because neither node can fit the CPU request.It runs by combining free CPU across both nodes.It reserves 4000m because placement uses limits.

QUESTION 6 · single-choice1 point

You delete a Pod owned by the lab Deployment while its desired replicas remain two. What should you observe after reconciliation?

Choose one answer.

You delete a Pod owned by the lab Deployment while its desired replicas remain two. What should you observe after reconciliation?A replacement Pod appears with a new UID.The deleted Pod returns with the same UID.Deleting the child also deletes its ReplicaSet.

QUESTION 7 · code-analysis2 points

The lab uses restartPolicy Always. Its synthetic container working set is 192 Mi. What does this resource fragment demonstrate?

Choose one answer.

resources:
  requests:
    cpu: 500m
    memory: 64Mi
  limits:
    cpu: 1000m
    memory: 128Mi

The lab uses restartPolicy Always. Its synthetic container working set is 192 Mi. What does this resource fragment demonstrate?The container is OOM-killed and may restart inside the same Pod.The container stops using memory at exactly its 64Mi request.The same Pod moves to another node to escape the limit.