Prometheus Monitoring

Prometheus Monitoring

Overview

Prometheus provides metrics collection and alerting for comprehensive application monitoring. This skill should be invoked when collecting application metrics, monitoring performance, or setting up observability.

Core Principles

• Metric Types: Counters, Gauges, Histograms, Summaries
• Pull Model: Prometheus scrapes /metrics endpoints
• Labels: Use labels for dimensional data
• Alerts: Define alert rules for anomalies

Preparation Checklist

• Set up Prometheus server
• Choose client library for your language
• Plan metrics and labels
• Design alerting rules

Step-by-Step Process

1. Add Library: Install Prometheus client
2. Define Metrics: Create counters, gauges, histograms
3. Expose: Implement /metrics endpoint
4. Configure: Set up Prometheus scrape config
5. Visualize: Create Grafana dashboards
6. Alert: Define alerting rules

Do's and Don'ts

• ✅ Do use appropriate metric types
• ✅ Do add meaningful labels
• ✅ Do set appropriate histogram buckets
• ❌ Don't use high-cardinality labels
• ❌ Don't expose sensitive data
• ❌ Don't forget to instrument code

Anti-Patterns

• Cardinality Explosion: Too many label values
• Missing Metrics: Not instrumenting critical paths
• No Alerts: Setting up Prometheus without alerting
• Sensitive Exposure: Exposing internal metrics