Correct Answer A, D
In order to avoid Single Point of failures of any kind, you have to have
- workers nodes with auto-healing and auto-scaling
- a backup master node (regional cluster).
Nodes execute the workloads run on the cluster. Nodes are VMs that run containers configured to run an application. Nodes are primarily controlled by the cluster master, but some commands can be run manually. The nodes run an agent called kubelet, which is the service that communicates with the cluster master.