We need to have a very clear and precise way of simulating pod density from the node point of view. 

What we need in a test suite -

1. Launch with our choice of workload
2. Gradually increase the pod density till we start seeing the failures on the node.
3. Notice the failures 
4. Gather the logs and exit

Success Criteria:

• Use kubeburner to make a node unschedulable or not ready, and then reboot the node to gather logs
  • AWS: did not go not ready
  • Azure: still pending
• Reproducible by other team members (google doc or github repo)