Metrics

Ricochet captures metrics that are exposed via a Prometheus server. The metrics are served at /metrics on port 6189 by default.

For example, to access the metrics from localhost:

curl http://localhost:6189/metrics

These metrics are on a separate port than ricochet so that they can be kept on an internal network.

Configuration

Set the RICOCHET_PROMETHEUS_PORT environment variable to serve metrics on a different port.

RICOCHET_PROMETHEUS_PORT=9100

Restart ricochet to apply the change.

Scrape configuration

Add ricochet as a target in prometheus.yml:

scrape_configs:
  - job_name: ricochet
    static_configs:
      - targets: ["localhost:6189"]

Counters and histograms

Metric	Type	Labels	Description
ricochet_requests_total	Counter	type, content_id, status, method	Requests served.
ricochet_connections_total	Counter	type, content_id	Connections opened.
ricochet_invocations_started_total	Counter	content_id	Invocations started.
ricochet_invocations_completed_total	Counter	content_id, status	Invocations completed.
ricochet_invocation_duration_seconds	Histogram	content_id	Time an invocation took to complete.
ricochet_app_instances_started_total	Counter	content_id	App instances started.
ricochet_app_instances_stopped_total	Counter	content_id	App instances stopped.

Gauges

Metric	Labels	Description
ricochet_active_processes	type, content_id	Processes currently running.
ricochet_connections	type, content_id	Connections currently open.
ricochet_memory_bytes	type, content_id	Memory currently in use.
ricochet_peak_memory_bytes	type, content_id	Highest memory use observed.
ricochet_longest_active_seconds	type, content_id	Age of the longest running process.
ricochet_last_invocation_timestamp_seconds	content_id	Unix time of the most recent invocation.