Skip to content

Verdaccio Plugin which exposes an OpenMetrics/Prometheus endpoint with health and traffic metrics

License

Notifications You must be signed in to change notification settings

freight-hub/verdaccio-openmetrics

Repository files navigation

verdaccio-openmetrics

Verdaccio plugin exposing an OpenMetrics/Prometheus endpoint with health and traffic metrics

This plugin, when installed and loaded, serves Prometheus/OpenMetrics metrics at a known path. This can then be scraped by your Prometheus installation, Datadog agent, etc.

By default, a second HTTP listener is used to keep metrics internal to your infrastructure more easily. The default metrics port is 9090. So, metrics will be available at :9090/metrics.

If you instead want to have your metrics available via Verdaccio's main API, set metrics_on_main: true and access /-/metrics on Verdaccio. This will disable the second HTTP listener unless you also pass metrics_port explicitly.

The primary metrics exposed are HTTP response latencies by request method and response status code. There's also an option to collect NodeJS runtime metrics (the defaults from the prom-client package). A further option will be to infrequently collect statistics about the database, however it's not clear yet what will be interesting to expose there.

config

If no extra config is given, HTTP request metrics will be exposed at :9090/metrics.

middlewares:
  openmetrics:
    enabled: true

If you want additional metrics, for example runtime and database metrics:

middlewares:
  openmetrics:
    enabled: true
    collect_runtime: true
    collect_database: true

All config keys and default values:

interface MetricsConfig {
  metrics_port: 9090; // Exposes metrics at /metrics on this alternate port
  metrics_on_main: false; // Exposes metrics at /-/metrics on Verdaccio
  default_labels: {}; // key/value tags attached to every metric

  collect_http: true; // Latency metrics for Verdaccio's API, by method and status code
  collect_database: false; // WIP metrics about Verdaccio's database, updated infrequently
  collect_up: false; // a fixed gauge of '1'
  collect_runtime: false; // the default Prometheus metrics for NodeJS processes
}

development

See the verdaccio contributing guide for instructions setting up your development environment. Note that yarn is being used in place of npm in this plugin. Once you have completed that, use the following yarn tasks.

  • yarn build

    Build a distributable archive

  • yarn test

    Run unit test

For more information about any of these commands run yarn ${task} -- --help.