Ioannis Angelakopoulos:
There is not a single highly available cluster in the pool of 6 available clusters during the exam.
So, is the topic: Manage a highly-available Kubernetes cluster relevant?
Ravi Shanker:
Possibly highly available and highly redundant are taken as separate design criteria. In highly available cluster there is always few minutes downtime from all documentation of cluster setup I have read so far. In this case the worker nodes are highly available. Moreover k8s gives flexibility in terms of keeping pods in different states in terms of availability.
Ioannis Angelakopoulos:
To have a highly-available cluster, the controlplane nodes (stacked or not) should be highly available.
The worker nodes alone are not enough to consider a cluster highly-available.
As i see it the topic is either mistakenly added in the curriculum, or there exists a question to create a highly available stacked cluster by converting the 2 worker nodes to controlplane, and finally having a cluster of 3 stacked controlplane nodes and removing the taints so you can have workloads in the cluster. 