Prometheus metrics

MinIO Sidekick exposes Prometheus-compatible metrics for monitoring load balancer performance and backend health.

Sidekick serves metrics at /.prometheus/metrics on the main listening port:

curl http://localhost:8080/.prometheus/metrics

The endpoint requires no authentication and returns metrics in Prometheus text format.

Sidekick exports these metrics for each backend:

• sidekick_requests_total - Total number of requests forwarded to the backend
• sidekick_errors_total - Total number of failed requests to the backend
• sidekick_rx_bytes_total - Total bytes received from the backend
• sidekick_tx_bytes_total - Total bytes sent to the backend

All metrics include labels identifying the backend endpoint.

Configure Prometheus to scrape the load balancer endpoint:

scrape_configs:
  - job_name: minio-sidekick
    metrics_path: /.prometheus/metrics
    static_configs:
      - targets:
          - localhost:8080

The default scrape interval of 15 seconds provides sufficient granularity for most deployments. Adjust the interval based on your monitoring requirements.

Monitor request distribution across backends by comparing sidekick_requests_total values. Large differences indicate uneven load distribution. Track error rates using the ratio of sidekick_errors_total to sidekick_requests_total. High error rates may indicate backend health issues.

Network throughput monitoring uses sidekick_rx_bytes_total and sidekick_tx_bytes_total. Calculate rates using Prometheus rate functions to identify bandwidth patterns and potential bottlenecks.