lazydns

Collector Plugin

The metrics_collector plugin counts DNS queries and accumulates latency information. When built with the metrics feature it also provides a Prometheus-backed collector (prom_metrics_collector) that registers metrics in a Prometheus Registry.

This page documents the in-process collector and the optional Prometheus integration.

What it provides

• A simple, test-friendly MetricsCollectorPlugin that maintains a shared counter and accumulated latency.
• When the metrics feature is enabled, a PromMetricsCollectorPlugin (it’s little weird that metrics enable prometheus, but oh well) registers Prometheus metrics (counters, gauge, histogram) and exposes more detailed telemetry.
• Exec-style quick-setup helpers are provided for simple instantiation in runtime configs.

Behavior

• Each executed plugin invocation increments a shared query counter.
• If query_latency_ms metadata is present (f64), the collector aggregates latency (sums) and exposes average latency via helper methods.
• The Prometheus variant records counters, error counters, a thread/activity gauge, and a latency histogram.

Configuration

The basic collector does not require configuration beyond construction. For runtime quick-setup the exec prefix metrics_collector is recognized.

YAML example (exec quick setup):

plugins:
  - exec: metrics_collector:

Prometheus-backed collector (requires metrics feature and a way to expose /metrics):

plugins:
  - exec: prom_metrics_collector:name=my_dns_server

The prom_metrics_collector quick-setup accepts name=<metric_name> to scope/register metrics under a unique name.

Prometheus integration

• When compiled with the metrics feature, PromMetricsCollectorPlugin registers metrics with a prometheus::Registry and provides the same execution behavior while exporting Prometheus metrics.
• Use the name=<metric_name> quick-setup to namespace metrics and avoid collisions.

Metadata compatibility

• The collector reads query_latency_ms (f64) metadata to observe latency.
• The Prometheus collector also checks query_error (bool) metadata to increment an error counter.

Troubleshooting

• If counts are not increasing: ensure the plugin instance is actually attached to the running pipeline and execute() is being called.
• For Prometheus metrics not appearing: confirm the metrics feature is enabled, the collector is registered with the correct registry, and the /metrics exposure endpoint is configured.

Best practices

• Use the in-process collector in tests and lightweight deployments.
• Use the Prometheus-backed collector for production telemetry; provide a stable name to avoid metric name collisions.

This site is open source. Improve this page.