infra-prometheus-optimization

What problem does it solve?

Prometheus setups degrade over time due to high cardinality, inefficient queries, overloaded remote write, and noisy alerting, causing storage bloat, performance issues, and operational fatigue.

Core Features & Use Cases

• Cardinality explosion diagnosis to identify metrics and label pairs with excessive series growth using TSDB status endpoints, then verify impact with PromQL.
• Recording rules using level:metric:operations naming to precompute expensive dashboard queries and standardize metric computation for faster panels and consistent alert logic.
• AlertManager tuning for routing and inhibition using routing trees to send critical incidents to the right receivers, apply inhibition to prevent alert storms, and use absent() to detect missing scrape targets.
• Storage and ingest efficiency controls via scrape interval tuning, relabeling/metric relabeling to drop expensive labels and metrics, and remote write configuration for Prometheus → Thanos/Mimir long-term retention.

Quick Start

Use the infra-prometheus-optimization skill to diagnose high-cardinality series, add recording rules for your slow dashboards, and tune remote write and alert routing based on your current Prometheus and AlertManager configuration.