The Two-Hour Game Plan and the Retake Rule

Why this matters

Every task carries a weight and the weights are published per domain, so the exam tells you where the points live. Candidates who work question one to last in order donate time to low-weight tasks and starve the heavy ones. A plan made before the exam converts the same knowledge into more points with zero extra studying.

Concepts

First pass, harvest: skim all tasks, solve the ones you can finish in under five minutes each, and flag the rest. This banks guaranteed points early and, more importantly, shows you the full exam while your mind is fresh; later tasks often reveal context (namespaces, contexts, cluster quirks) that unblocks earlier ones.

Second pass, weight order: work flagged tasks heaviest first. A task worth double deserves double the time, and a task you half-finish still earns partial credit on most graders, so never leave a heavy task untouched while polishing a light one.

The stuck rule: fifteen minutes on one task with no progress means flag and move. Stuck time is the only unrecoverable resource in the exam; a fresh look after two other tasks solves more blocks than grinding does.

The retake rule: the program includes one free retake, so a first attempt is reconnaissance with upside, not a final verdict. That changes risk calibration: attempt every task, guess the field instead of skipping, and treat an unfamiliar cluster quirk as information for the plan, not a disaster.

Worked example

A 120-minute budget for a typical weight spread:

minutes 0-10:   skim all tasks, solve the trivial ones, flag the rest
minutes 10-80:  flagged tasks in weight order (heaviest first)
minutes 80-105: remaining flags, partial attempts everywhere
minutes 105-120: verification sweep (below), no new tasks

The sweep re-runs one read per task: kubectl get on what you created, describe on what you fixed, endpoint checks on what you exposed. A task that looks done but is not is worth zero; fifteen minutes of verification routinely rescues a full task worth of points.

Exam notes

Context switches (clusters, namespaces, users) are where points silently die: every task starts with confirming kubectl config current-context and namespace, every task ends with the verification read. Keep water and a clock visible; the exam timer is authoritative but a personal countdown keeps the phase changes honest.