Telemetry metrics definitions for Forge pipelines.

This module defines metric specifications that can be consumed by various telemetry reporters (StatsD, Prometheus, etc.) using the TelemetryMetrics library pattern.

Metric Categories

Counters

• Pipeline completions (by outcome)
• Stage executions (by stage, outcome)
• Retry attempts (by stage)
• DLQ enqueues (by stage, error)
• Measurement cache hits/misses

Distributions (Histograms)

• Pipeline duration
• Stage latency (by stage)
• Measurement computation time
• Storage operation latency
• Artifact upload/download sizes

Usage with Reporters

This module provides metric definitions that can be used with different telemetry reporters. Since telemetry_metrics is an optional dependency, these are provided as plain maps that reporters can adapt.

Example: Custom Reporter

defmodule MyApp.TelemetryReporter do
  def attach do
    metrics = Forge.Telemetry.Metrics.metrics()

    Enum.each(metrics, fn metric ->
      :telemetry.attach(
        "my-reporter-#{metric.name}",
        metric.event_name,
        &handle_event/4,
        metric
      )
    end)
  end

  defp handle_event(event, measurements, metadata, metric) do
    # Report to your monitoring system
  end
end

Example: StatsD Reporter

defmodule Forge.Telemetry.StatsDReporter do
  def attach do
    metrics = Forge.Telemetry.Metrics.metrics()

    Enum.each(metrics, fn metric ->
      case metric.type do
        :counter ->
          :telemetry.attach(...)

        :distribution ->
          :telemetry.attach(...)
      end
    end)
  end
end