A check fails or grading seems wrong

Klastro grades your work by checking the live state of your environment, not the commands you typed. A check passes when the resources the task asks for exist and are in the expected state at the moment the check runs. For how checks work, see Tasks and checks.

Each problem below is laid out as Symptom, Cause and Fix.

“Not yet — try again”

Symptom. In a tutorial, Check this step ends on Not yet — try again. In a challenge’s results, a task shows Not passed.

Cause. The environment does not match what the task asks for yet. The most common reasons:

• A name differs: the resource, namespace, label or file name is not exactly the one in the task. Names are case-sensitive.
• The wrong namespace: the resource was created in default instead of the namespace the task names.
• Not ready yet: you checked while pods were still starting, an image was still downloading, or a rollout was in progress.
• The wrong machine: in labs with several machines, the change was made on another machine than the one the task targets.
• Partly done: the task has several requirements and one is missing (a port, a replica count, a label).

Fix.

1. Reread the task word by word and compare every name with what you created, for example with kubectl get <kind> -n <namespace> and kubectl describe.
2. Wait until the resources are ready (kubectl get pods -w, kubectl rollout status), then select Check again.
3. If you are stuck, open the hints. In a tutorial, hints are free. In a challenge, revealing a hint costs points, shown next to it.

The check stays on “Checking…”

Symptom. The step shows Checking…, or the results show Grading… and Validating your work against the live environment…, for a long time.

Cause. Checks run against your live environment and can take a few moments, especially in labs with several machines.

Fix. Wait. The result appears on its own; you do not need to select the button again. If it never completes, report a problem from the lab.

My work is right but the check fails

Symptom. You are confident the environment matches the task, and the check still fails after Check again.

Cause. It can be a difference you have not spotted, or, more rarely, a problem with the check itself.

Fix. Check the list above once more, then report a problem from the lab with the category Lab environment. Include the task name, what you created and the output of the kubectl commands that show it. The lab’s details are attached, so support can examine your environment while it is still running: do not end the session before you have reported it.

A passed step shows as not passed after I changed something

Symptom. A step you had passed is no longer shown as complete after you checked again.

Cause. Each check reflects the environment at the time it runs. If you deleted or changed what an earlier task created, a new check of that task sees the new state.

Fix. Restore what the task asked for and check again.

In an assessment

During an assessment, your tasks are also graded from the state of your environment. You can move freely between scenarios, but you cannot return to the lab once you select Complete Assessment, so make sure every task is done before you confirm. See Assessment mode.

Caution

Do not end or delete a lab before its grading has finished. Grading reads your live environment: once the environment is destroyed, there is nothing left to check.