this is related to the moment in time after we will have merged #246
some of the kps-based dashboards that are generated contain incovenient variables for filters, for instance node exporter contains variable for instance which is just an IP:port of node-exporter target, which is really annoying to figure out each time
we should add some modifications that would leverage mixins jsonnet mechanisms and:

• add a variable for nodename that filters for node labels in the selected cluster
• modify all the queries in the dashboards to use node=$node instead of instance=$instance

the way I see it is like this:

kps charts generate dashboards based on this
https://github.com/prometheus-community/helm-charts/blob/main/charts/kube-prometheus-stack/hack/sync_grafana_dashboards.py
for node exporter it pulls prometheus-operator/kube-prometheus dashboard definitions that in turn are generated with make generate in the repo and include mixins of various components such as node-exporter

https://github.com/prometheus/node_exporter/tree/master/docs/node-mixin

there you can see that it is (theoretically) possible to override some of the fields and values there, either when jsonnet imports node-mixin and renders a dashboard json or by (again, theoretically) using kustomizations within prometheus-operator/kube-prometheus make generate