Metrics

In the tables

Current CPU and memory come from metrics-server and appear in the rows themselves.

History

CPU and memory over time come from Prometheus, auto-discovered and queried through the apiserver proxy rather than needing a route of its own.

spinoza --prometheus monitoring/prometheus-server:80

With no Prometheus to ask, Spinoza samples metrics-server itself every 15 seconds while the panel is open. Each series is labelled with where it came from and, when Spinoza sampled it, when sampling began, so a short window never reads as history the cluster kept.

Reserved against used

What every namespace and workload asks the scheduler for, against what it actually uses, ranked by what could be given back.

Usage comes from Prometheus over a window where there is one, and from metrics-server otherwise. Where nothing measured a workload, the view says so rather than showing a zero.