LAB 4.2 may break k8s cluster if you try to use more resources for the stress app
If you ask the stress app to consume exactly the same amount of resources as your master - should be allowed to be scheduled workload on it - node has, it will break the cluster.
Maybe OOM killer or someone else kills the api.
Final state of this experiment is no api server running on my master node
Is it desired state/behavior of k8s cluster ?