Mock exams
Drills build one skill at a time. A dedicated mock tests whether you can put those skills together under one clock.
kubefit packs cka # download signed drills and the separate exam form
kubefit test cka # 17 original questions, 120 minutes
kubefit test ckad # same format, a different track-specific exam
Each of the eight tracks has one dedicated form containing 17 questions, separate from its 25 drills. Each form totals 100 points. These are original KubeFit practice questions, not recalled certification questions. The two-hour format is a KubeFit benchmark; official exams can have different formats, durations and permitted resources. Operations is a supplemental training track, not a separate certification.
Before the clock starts
KubeFit verifies the exam's Ed25519 signature, creates a fresh local environment, installs its prerequisites and seeds all questions. Setup time is outside the exam clock. Kubernetes exams use dedicated kubefit-exam-<track> kind clusters with Cilium; Istio also requires istioctl. LFCS uses its own kubefit-exam-lfcs Lima VM. Your drill cluster and drill VM are separate.
Bring Docker, kind, kubectl, Helm, Python 3 and Bash for Kubernetes exams. LFCS needs Lima, and Python 3 and Bash as well. kubefit doctor checks both from CLI v0.1.9. Initial setup needs an internet connection for images and controllers. Allow room for a three-node cluster with Cilium, plus Istio for the ICA exam. Avoid running several environments together on a memory-constrained machine.
Work in any order
All questions are available when the two-hour clock starts:
kubefit task # list all 17 questions and their points
kubefit task 6 # display question 6 and enter its working directory
kubefit clock # remaining time and deadline
kubefit submit # finish early and grade the complete exam
The exam shell is a clean Bash started without your own profile, so your aliases and prompt are absent. Its prompt reads exam <dir> ❯, the remaining time prints before every prompt, and it opens on question 1. task N changes to that question's directory, sets NS, QUESTION and INPUTS, and switches the current kubectl context to that namespace. kubectl -n "$NS" … still works and is the safer habit. Each question states its namespace, required outcome and any supplied inputs. Paths beginning inputs/ and answers/ are relative to that question's directory. Return to any question with task N; changing questions does not reset anything.
task, clock, submit and escape also work from a second terminal. Every other kubefit command except login, logout, sync, version, help and report is refused until the exam ends, including note, packs, doctor and end.
Use your own verification commands. Interim grade, hints, solutions, Coach Caz assistance and pause/resume are unavailable during the exam. The clock continues while you read, change questions or put the laptop to sleep. Check your certification provider's current allowed-resources rules; links in the guides are navigation practice, not a promise that every linked site is allowed in every exam.
When time expires
At 120 minutes, KubeFit stops the exam shell and its work processes, then grades all 17 questions, including unfinished ones. You do not need to press Enter. The same final grading runs when you enter kubefit submit or exit the shell early. If the computer was asleep, grading runs when it wakes; the recorded deadline is unchanged.
Checks inspect the final cluster/VM state or a specifically requested artifact. Some security and platform questions explicitly use offline evidence or configuration files; those checks do not claim to verify a running external service. Controllers may need a brief reconciliation period after work stops. Grading time is not extra working time.
Points within a question are split equally across its published outcome checks. The report shows each check, points earned, the total out of 100, and whether you submitted or timed out. An incomplete checker cannot award points, and if any checker could not run the report says Grading incomplete: no overall score instead of a total. A score is practice feedback, not a prediction or an official passing threshold.
The report is saved locally, printed automatically, and synced to your account when online. Use kubefit report <track> or the site's Reports page to revisit it. Exam results do not create drill reps or cold passes.
Leave without a grade
kubefit escape ends the entire exam ungraded. It does not record partial drill reps. A process interruption also leaves ungraded evidence; run kubefit test --recover if the supervisor stopped unexpectedly. Recovery closes that attempt—it does not resume a stopped clock or pretend work after an interruption was supervised.
kubefit end removes KubeFit's local exam and drill environments. Reports remain in your local history. Starting the same exam again creates a fresh environment and reuses that track's fixed form. Repeated scores show practice consistency, not performance on unseen questions.
Caz: Seventeen questions. One clock. Save the victory lap until the report exists.