Context:

As we start receiving metrics consistently in OCM environments and we are creating SLOs dashboards that can consume data from any data source Prod/stage/CI we also want to revisit how we are sending metrics and make sure we are doing it int the most effective way. We have some wonky data coming through in prod atm.

DoD:

Atm we have high frequency reconciliation loop where we constantly review the over all state of the world by looping over all clusters.

We should review this approach and record metrics/events as it happens directly in the controllers/reconcile loop only once and not repeatedly in a loop when possible for each specific metric.