Description of problem:
When running node-density (245 pods/node) on a 120 node cluster, we see that there is a huge spike (~22s) in Avg pod-latency. When the spike occurs we see all the ovnkube-master pods go through a restart.
The restart happens because of (ovnkube-master pods)
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Run node-density on a 120 node cluster
Spike observed in pod-latency graph ~22s
Steady pod-latency graph ~4s