Prometheus metrics

LokiLoggerModule can expose a Prometheus /metrics endpoint with both Node.js system metrics and per‑route HTTP metrics suitable for production dashboards.

Enabling metrics

In your module registration:

LokiLoggerModule.register({
  serviceName: "backend",
  lokiHost: "http://loki:3100",
  enableMetrics: true,
  metricsPath: "/metrics", // optional, default
});

When enableMetrics is true, LokiLoggerModule.apply(app):

1. Creates a MetricsService with its own Prometheus Registry.
2. Calls collectDefaultMetrics from prom-client with { app, env } labels.
3. Mounts GET /metrics on the underlying Express instance.

Exposed metrics

The metrics implementation (see MetricsService) defines:

• http_request_duration_seconds – Histogram

  • Labels: method, route, status_code, app, env
  • Buckets: 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10
  • Populated from the middleware on each response.

• http_requests_total – Counter

  • Labels: method, route, status_code, app, env

• http_requests_in_flight – Gauge

  • Labels: app, env
  • Incremented on request start, decremented on finish.

In addition, collectDefaultMetrics exposes process‑level metrics such as:

• CPU usage
• Heap usage and GC stats
• Event loop lag
• Open file descriptors and handles

Scraping with Prometheus

In the provided stack, Prometheus is configured in:

• docker/services/prometheus/prometheus.yml
• docker/targets.yaml

prometheus.yml defines a nestjs-services job that reads targets from targets.yaml:

scrape_configs:
  - job_name: "nestjs-services"
    metrics_path: "/metrics"
    scrape_interval: 10s
    file_sd_configs:
      - files:
          - /etc/prometheus/targets.yaml
        refresh_interval: 30s

targets.yaml is where you add or change your services:

- targets: ["backend:3003"]
  labels:
    app: "backend"
    env: "production"
 
- targets: ["auth-service:3001"]
  labels:
    app: "auth-service"
    env: "production"

Prometheus hot‑reloads this file every 30 seconds; no restart is required.

Example Grafana queries

Once Prometheus is scraping your services, you can build dashboards (or use the included ones) with queries like:

• P95 latency per route

histogram_quantile(
  0.95,
  sum by (le, route) (
    rate(http_request_duration_seconds_bucket[5m])
  )
)

• Requests per second per service

sum by (app) (rate(http_requests_total[1m]))

• In‑flight requests by service

http_requests_in_flight

Production checklist

• Ensure each service sets a consistent env value (e.g. production, staging) in its Loki options so labels align across logs and metrics.
• Expose /metrics only on internal networks or behind authentication – the endpoint returns detailed system information.
• Keep PROMETHEUS_RETENTION in env.example aligned with how long you need metrics for SLO calculations and debugging.

Need help or want to support the project? Visit Support.