Prometheus Metrics: Pull Scraping, Time Series, And PromQL Basics

**Prometheus** scrapes HTTP **metrics** endpoints on a pull interval. Metrics are named time series with labels (`http_requests_total{method="GET",status="200"}`). Documentation: `prometheus.io/docs`.

Metric types include **counter** (monotonic), **gauge** (up/down), **histogram** and **summary** (latency distributions). Histograms enable quantile estimation in PromQL with `histogram_quantile`.

**Alertmanager** routes firing alerts to paging and chat systems with grouping and silences. Recording rules precompute expensive queries.

Prometheus is not long-term log storage; pair with Loki or ELK for logs and Tempo/Jaeger for traces in a full observability stack.