I just had a chat with @mgoodness regarding the state and future of the kube-prometheus helm charts, and we wanted to share what our vision is regarding them and we invite anyone who wants to help on this to participate.

We imagine the general structure of the charts to be like this:

The kube-prometheus chart (this is mostly a "meta" chart that includes multiple other charts and adds the Prometheus alerting/recording rules, grafana dashboards as well as service-monitors.

It's dependencies are:

• Prometheus chart (with opinionated configuration tailored for cluster monitoring, and to work with the alertmanager instance shipped)
• Alertmanager chart
• Prometheus Operator chart (because it's a dependency of the prometheus and alertmanager chart)
• grafana chart (without dashboards, but preconfigured with the k8s prometheus as datasource) (this one is already actively maintained by @weiwei04)
• kube-state-metrics chart
• node-exporter chart

Something that we had previously done in the kube-prometheus charts is also consider the choice of how the web UIs of the above components are exposed, we'd like to move away from this and leave this completely up to the user.

If anyone is willing to work on any of these charts, contributions are highly appreciated and very welcome. The individual charts that are a dependency of the kube-prometheus chart should be relatively low maintenance as the upstream projects are not released that often.