Skip to content

Latest commit

 

History

History
26 lines (17 loc) · 2.46 KB

File metadata and controls

26 lines (17 loc) · 2.46 KB

Observability

Audience: SRE, Platform engineer

Every component exposes Prometheus metrics from the standard controller-runtime metrics server, so built-in metrics (reconcile latency, work queue depth, etc.) are emitted automatically alongside the custom metrics documented in this guide. The serving posture differs by component:

For SLO targets associated with these metrics, see Appendix A — Capacity Targets & SLOs.


This guide

Observability is split by surface so you can read just the part you need without loading the whole thing:

Page What's here
Metrics reference Every metric the GMC, AGC, and proxy emit — the full catalogue, the scale-set acquisition tier, proxy metrics, the kubectl status columns and owner-label selectors, the label-cardinality guardrail, and the Q205 metric/span renames.
Accessing metrics (scraping setup) How to scrape: ad-hoc port-forward, the Prometheus Operator wiring, the install-time NetworkPolicy prerequisites, TLS verification for the GMC manager, and per-tenant AGC/proxy mTLS.
Alerting & SLOs The symptom → metric map, the ready-to-apply PrometheusRule alert rules, and the SLO recording rules.
Grafana dashboards The bundled tenant and platform dashboards, panel by panel, and their template variables.
Logging & tracing Structured JSON logging, per-tenant log levels, debug diagnostics for otherwise-silent paths, and OpenTelemetry tracing (AGC).

The alert rules and both dashboards also ship as code under deploy/monitoring/ — a PrometheusRule plus two Grafana dashboards — so prefer applying those over hand-copying the YAML.