pg_exporter

Project Overview

PG Exporter is an advanced PostgreSQL & pgBouncer metrics exporter for Prometheus, bringing ultimate monitoring experience with declarative config, dynamic planning, and customizable collectors.

Key Features

• Highly Customizable: Define almost all metrics through declarative YAML configs
• Full Coverage: Monitor PostgreSQL (10-19+) and pgBouncer (1.8-1.25+) in a single exporter
• Fine-grained Control: Configure timeout, caching, skip conditions, and fatality per collector
• Dynamic Planning: Define multiple query branches based on different conditions
• Production-Ready: Battle-tested in real-world environments across 12K+ cores for 6+ years
• Auto-discovery: Automatically discover and monitor multiple databases within an instance
• Health Check APIs: Comprehensive HTTP endpoints for service health and traffic routing
• Extension Support: Supports timescaledb, citus, pg_stat_statements, pg_wait_sampling, and more

Target Audience

• PostgreSQL database administrators and operations engineers
• Teams using Prometheus for PostgreSQL monitoring
• Users who need deep PostgreSQL observability

Quick Start

1. Download RPM/DEB/Tarball from GitHub Releases or Pigsty's YUM/APT Infra Repo
2. Pass the postgres/pgbouncer URL via env or argument: PG_EXPORTER_URL='postgres://user:pass@host:port/postgres' pg_exporter
3. Access metrics at http://localhost:9630/metrics
4. All other metrics are defined in the pg_exporter.yml config file

Core Advantages

• 600+ metrics, ~3K time series per instance, covering everything for PostgreSQL observability
• Zero-config, dynamic planning, production-grade reliability
• Grafana monitoring dashboards available

System Requirements

• OS: Cross-platform
• Architectures: x86_64, arm64