Why is Google Compute Engine CPU usage higher than expected?
I'm following the container engine walkthrough and I see that my VM Instance has a min CPU usage of ~80%. However, if I ssh into this box and run 'top' I see a much lower utilization. Can someone explain this to me as I must be missing something simple. Thank you. 
Similar Questions
1 Answer
There is a known (benevolent) bug in the ingress controller that is automatically added to your cluster in Kubernetes 1.1.1. If you are not using the controller, you can scale the number of replicas to zero:
kubectl scale rc l7-lb-controller --namespace=kube-system --replicas=0which should make your CPU usage go back to a normal level.
The ingress controller isn't doing any harm (other than affecting monitoring metrics) and will be automatically nice'd by the kernel if you run other pods on the same node (so it isn't affecting performance of your cluster).
This bug will be fixed in the upcoming 1.1.2 release of Kubernetes.