hossein bazrafkan:
Hi Guys,
I attended to CKA exam but I forgot to switch cluster before doing etcd backup restore so I removed my answers to many question, and I failed. Do you know they will accept my excuse? I mean the last state of cluster will be used for appraisal?
Alistair Mackay:
Unfortunately I can’t see that not executing the instructions given in the question would be accepted as an excuse.
Kind of like applying something to the wrong cluster at work and causing a major incident.
Re-credits are given for failure in the exam infrastructure, not user error.
M A:
@hossein bazrafkan mostly etcd has one or two questions, did you mean you did not choose the correct clusters for many questions?
hossein bazrafkan:
@M A
I mean I restored the etcd backup in wrong cluster, so I cleaned answers of other questions
M A:
ahh that should not happen, questions should be independent of each other , i would have thought
Alistair Mackay:
There are 6 clusters as I remember which are shared between all the questions. They almost certainly reserve one for etcd backup/restore and or kube version upgrade, so that if you blow that cluster it doesn’t affect other questions.
However if you restore the wrong cluster, on which you have answered other questions, then the restore will wipe out those answers. By definition, it’s restored the cluster to a state that doesn’t have your answers.
hossein bazrafkan:
that is happened.
And my question is if I request and pay, they will watch screen records and count other questions score or the final state of clusters are considered as answers?
Alistair Mackay:
The score is done by marking scripts which would use kubectl and curl to check the state of each cluster for the requirements of the questions. The score is calculated from what those scripts find.
If you answered questions on cluster A, you were meant to restore cluster B, but in fact restored cluster A, everything you did on cluster A is gone, so the marking script will score zero for everything that should have been present on cluster A, but now isn’t. It will look like you never attempted the questions.
Alistair Mackay:
I would be extremely surprised if they were to re-credit you, because what you actually did was to get the restore question incorrect - in the worst possible way.
Alistair Mackay:
If you’re confident that what you did do was correct (before you wiped it out), then book the retake, and double check the cluster you set before doing either of the 2 big destructive questions (upgrade and restore).
Double check anyway for all questions. Copy the kubectl config from each question and paste to the terminal before starting the question - even if you think it’s the same cluster as the previous question - just to be 100%
hossein bazrafkan:
I did it multiple time, maybe I did stupid mistakes. I will retake, thanks for your time and effort
M A:
@Alistair Mackay if we have multiple terminal windows open, do we need to set the context on all of them?
hossein bazrafkan:
I think if they are in the same node, no
Alistair Mackay:
If you’re on the same node (student-node, exam-node or whatever they call it), remember that the context is set by the kubeconfig file in ~/.kube/config
All terminals logged in as the same usere on the same node see the same kubeconfig - thus the context is set for all.
Just do kubectl get pods in 2 terminals after setting the context. You will see the same pods (if there are pods in the current namespace)
M A:
understood, thank you, so probably a good idea to exit back to student/exam node after each question then.
Alistair Mackay:
It is what they suggest you do, if you have had to SSH to any worker node