Skip to content

observability.md

Latest commit

 

History

History
64 lines (45 loc) · 2.24 KB

observability.md

File metadata and controls

64 lines (45 loc) · 2.24 KB

Observability

The project includes a complete observability stack for monitoring, logging, and performance analysis.

Metrics

The system automatically collects the following metric categories:

  • HTTP metrics

    • Total number of HTTP requests by method and endpoint
    • Request latency (p50, p95, p99)
    • Error count by type (4xx, 5xx)

  • Database metrics

    • Number of database queries
    • Database query latency
    • Connection error count
    • Database connection status

  • Frontend metrics (Web Vitals)

    • LCP (Largest Contentful Paint) - main content load time
    • INP (Interaction to Next Paint) - interaction response time
    • CLS (Cumulative Layout Shift) - layout stability
    • FCP (First Contentful Paint) - first content render time
    • TTFB (Time to First Byte) - time to first byte

  • PostgreSQL metrics

    • Number of active connections
    • Database and table sizes
    • Query and transaction statistics
    • Replication metrics (if configured)

All metrics are available through Prometheus at http://localhost:9090 and visualized in Grafana.

Logs

Centralized logging is implemented through Grafana Alloy and Loki:

  • Automatic log collection from all containers via Docker labels
  • Storage in Loki with configured retention
  • Search and analysis through Grafana Explore
  • Structured logs with metadata (timestamp, level, service)

Log access:

Grafana Dashboards

After startup, a ready-made dashboard DevOps Demo Dashboard is automatically available, which includes:

  • General API statistics (requests per minute, latency, errors)
  • Database metrics (active connections, queries, latency)
  • Web Vitals metrics from frontend
  • PostgreSQL metrics (DB size, active transactions)
  • Trend graphs for various time periods

The dashboard is automatically picked up through Grafana provisioning and is available immediately after startup.

Alerting

Prometheus is configured with basic SLO rules (Service Level Objectives), defined in observability/prometheus_slo_rules.yml. Detailed information about SLO is available in SLO.md.