Byi override yaml update

[bin3377]

update yaml files for:

• fix the changed name/job in metrics
• add cluster label
• remove cadvisor source since it's combined into kubelet

[frankreno]

missing the t in cluster

[bin3377]

@ggarg2906sumo this change should fix all missing time series. These things need to be noticed:

1. There is an incompatible between Prometheus-Operator chart and latest version of Kubernetes. So we need a workaround for install helm chart
   a) clone https://github.com/helm/charts/
   b) modify values.yaml in (path-to-charts-repo)/stable: find component: kube-controller-manager and replace to k8s-app: kube-controller-manager; find component: kube-scheduler and replace to k8s-app: kube-scheduler
   c) install chart with modified values.yaml with running following command in (path-to-charts-repo)/stable:
   helm install --namespace sumologic --name prometheus-operator ./prometheus-operator -f prometheus-operator/values.yaml -f (path-to-sumologic-kubernetes-collection)/deploy/helm/overrides.yaml --no-crd-hook

2. time series in cadvisor is combined into kubelet

3. some metrics name changed in node-exporter part

[frankreno]

why clone? why not set in helm overrides as it can override those value?

[frankreno]

I also think that due to the version change and use of many k8s versions in the wild, we should have an FAQ doc or something and just call this out specifically.

[bin3377]

I tried with make these changes into override.yaml, but it will not overwrite the original selector but add them as:

Name:              prometheus-operator-kube-controller-manager
...
Selector:          component=kube-controller-manager,k8s-app=kube-controller-manager

which will make the service is still not connected since the selector is actually working as AND

[frankreno]

Crap. I forgot about that bug in helm. Ok is what it is for the moment.

[bin3377]

@lei-sumo @samjsong