You need to prevent a node pool from running user workloads while still allowing kube-system pods. Which mechanism do you use?
- A Node selector
- B Pod disruption budget
- C Taint with effect NoSchedule and matching toleration in kube-system pods (correct answer)
- D Cordon the node
Shortcut: Tainting system nodes with CriticalAddonsOnly:NoSchedule (added automatically to system node pools) prevents regular workloads from scheduling. kube-system pods have the matching toleration.