Course contentsShow
← Cloud and Infrastructure
Lesson 1956 of 2,98942. Observability - MetricsFree lesson

Prometheus Configuration Basics

Configuring prometheus.yml with scrape configs, targets, intervals, and global settings for the monitoring server.

Prometheus Configuration Basics

What you'll learn: How to configure Prometheus to collect metrics from your services using the prometheus.yml configuration file.


Prometheus uses a single YAML configuration file—typically named prometheus.yml—to define what to monitor, how often to collect metrics, and where to find targets. This file is the control panel for your entire monitoring operation.

Core Configuration Structure

The configuration file has several key sections:

Global settings define defaults that apply to all scrape jobs:

global:
  scrape_interval: 15s      # How often to collect metrics
  evaluation_interval: 15s  # How often to evaluate rules
  external_labels:
    cluster: 'production'   # Labels attached to all time series

Scrape configs define individual monitoring jobs and their targets:

scrape_configs:
  - job_name: 'my-app'
    scrape_interval: 10s    # Override global setting
    static_configs:
      - targets: ['localhost:8080', 'localhost:8081']
        labels:
          env: 'prod'

Think of It This Way

Imagine Prometheus as a newspaper reporter with a beat. The global section sets their daily routine (visit sources every 15 minutes). Each scrape_config is a specific source—a city hall, police station, or school board. The targets are the exact addresses, and labels are tags to categorize the stories.

Key Elements

  • scrape_interval: Frequency of metric collection (per-job overrides global)
  • targets: List of host:port endpoints exposing metrics
  • job_name: Logical grouping automatically added as a label
  • labels: Custom dimensions attached to all metrics from these targets

Key Takeaway: The prometheus.yml file controls all monitoring behavior through global defaults and per-job scrape configurations, defining targets, intervals, and labels that structure your entire metrics collection strategy.