Metrics Scraping

• Prometheus scraping works by periodically sending HTTP requests to specific endpoints on the target systems (called targets) to collect metrics. This process is called scraping

Targets

• Prometheus monitors the targets
• Linux/Windows Server, Single Application, Apache Server, Databases, etc
• Each target has units, which expose the metrics
• CPU, memory/disk usage, exception counts, requests count, request duration, etc

Target Exposition and Metrics Endpoint

• Targets (e.g., applications, services, or exporters) expose metrics via an HTTP endpoint, typically in a format like /metrics (e.g., karpenter.kube-system.svc.cluster.local:8080/metrics)
• These metrics are usually in the Prometheus text-based exposition format

Prometheus Configuration

• In the prometheus configuration file prometheus.yml, you define the scrape job with

scrape_configs:
  - job_name: 'example'
    scrape_interval: 15s
    static_configs:
      - targets: ['localhost:8080']

• Prometheus sends an HTTP GET request to the /metrics endpoint of each target at the specified interval

Time-series database

• Prometheus processes the collected metrics and stores them in its time-series database for querying and alerting.
• The database can be queried via PromQL for visualization and alerting