Allow for whole-cluster scraping selectors from mimir-distributed Helm-Chart about mimir HOT 2 CLOSED

This pushes the responsibilities of the helm chart further than now. I think it makes the problem of configuring the agent harder - how do you define pod logs and service monitors to cover all pods however they may be configured? Do we allow arbitrary relabelling?

I think we can start with an opinionated setup which supports only one kind of deployment (only TCP http-metrics ports scrape all namespaces). But I can see this quickly increasing in scope as different users have different needs.

sounds related to #5860 too

Perhaps some middle ground is using the extraObjects in values.yaml to add a ServiceMonitor manifest which covers the pods you'd like to scrape and let the agents scrape them too. IIRC the only requirement would be on the labels of the ServiceMonitor - they should match the serviceMonitorSelector in the MetricsInstance created by the chart

from mimir.

I agree and we've since stopped exploring mimir as a viable options as well.
Due to, primarily, the lack of feature parity with Kube-Prometheus-stack.

from mimir.