You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At the moment, all KubeCost components are deployed into the shoot cluster (including its own Prometheus).
This can be improved by leveraging the Prometheus and Plutono deployed in the seed cluster. Since the cost-analyzer needs query access to Prometheus, it makes sense to deploy the cost-analyzer into the seed cluster too. I believe the only component that needs to run in the shoot cluster is the network cost daemonset, as this will analyze the traffic going in and out of pods on the shoot nodes.
A few things that need investigating:
are all the necessary metrics collected from node-exporter and cadvisor?
do all metrics have the necessary labels?
how can the kubecost manifests be split across seed/shoot while still make them easily updatable?
The text was updated successfully, but these errors were encountered:
At the moment, all KubeCost components are deployed into the shoot cluster (including its own Prometheus).
This can be improved by leveraging the Prometheus and Plutono deployed in the seed cluster. Since the cost-analyzer needs query access to Prometheus, it makes sense to deploy the cost-analyzer into the seed cluster too. I believe the only component that needs to run in the shoot cluster is the network cost daemonset, as this will analyze the traffic going in and out of pods on the shoot nodes.
A few things that need investigating:
The text was updated successfully, but these errors were encountered: