Metrics

liteio exposes a Prometheus-compatible metrics endpoint on the console port. Set --metrics-token to enable it; requests must carry the token as a bearer credential.

curl -H "Authorization: Bearer scrape-token" \
  http://localhost:9001/metrics

Scrape configuration

scrape_configs:
  - job_name: liteio
    bearer_token: scrape-token
    static_configs:
      - targets: [liteio.example.com:9001]
    metrics_path: /metrics

Metric catalog

Request metrics

Metric	Type	Description
liteio_s3_requests_total	counter	Total S3 requests, labeled api, bucket
liteio_s3_errors_total	counter	S3 errors, labeled api, code
liteio_s3_request_duration_seconds	histogram	Request latency, labeled api
liteio_s3_inflight_requests	gauge	Requests currently being processed
liteio_s3_bytes_in_total	counter	Upload bytes received
liteio_s3_bytes_out_total	counter	Download bytes sent

Cluster health metrics

Metric	Type	Description
liteio_cluster_drives_online	gauge	Drives currently reachable
liteio_cluster_drives_total	gauge	Total configured drives
liteio_cluster_sets_degraded	gauge	Erasure sets with at least one offline drive
liteio_cluster_sets_below_read_quorum	gauge	Sets that cannot serve reads
liteio_cluster_capacity_raw_bytes	gauge	Raw storage bytes across all drives
liteio_cluster_capacity_usable_bytes	gauge	Usable bytes after parity overhead
liteio_cluster_capacity_used_bytes	gauge	Bytes currently in use
liteio_cluster_heal_queue_depth	gauge	Objects queued for reactive healing

Go runtime metrics

Standard go_* and process_* metrics from the Go runtime are also exported: GC pause duration, heap size, goroutine count, CPU usage.

Example queries

Requests per second by API:

rate(liteio_s3_requests_total[5m])

p99 GET latency:

histogram_quantile(0.99, rate(liteio_s3_request_duration_seconds_bucket{api="GetObject"}[5m]))

Error rate:

rate(liteio_s3_errors_total[5m]) / rate(liteio_s3_requests_total[5m])

Capacity used fraction:

liteio_cluster_capacity_used_bytes / liteio_cluster_capacity_usable_bytes