Native Prometheus Integration with Nginx

Categorica has open-sourced nginx-stats-prometheus, an NGINX module that exposes Prometheus metrics directly from the server instead of running a separate exporter web service or sidecar container. The team argues that a separate process adds deployment complexity and leaves the metrics collector itself outside the monitoring system. The module offers a simpler option for deployments that prefer native collection, though it is not intended as a full replacement for the established NGINX Prometheus Exporter.

The module is built against NGINX source and requires a C99 compiler and supporting libraries; existing NGINX installations need compatibility and stub-status features enabled. After installation, a configuration directive exposes a chosen endpoint such as /metrics. Available measurements include process status and ID, HTTP request and response counts, traffic by status class, method and protocol, request duration histograms, and connection statistics. It also reports basic per-server and per-upstream totals after traffic reaches them. Collection starts at process launch, but shared memory defaults to 1 MB, supports at most 512 server or upstream directives, and may silently skip metrics if memory runs out. The project is MIT-licensed and accepts reviewed contributions, but offers no support guarantee. Its trade-offs are fewer metrics, a required rebuild when NGINX changes, and less testing and maturity.