Meters4s is a thin, functional wrapper around Micrometer designed to integrate with the Cats ecosystem. This allows for in-process metrics aggregation for counters, timers, gauges and distributions.

Add the following dependency to your build.sbt:

libraryDependencies += "io.github.keirlawson" %% "meters4s" % "<see latest version>"

Or for Cats Effect 2.x use the 0.4.x series.

You will likely also want to add the module corresponding to whichever monitoring system you want to report metrics to. All systems supported by micrometer can be used by brining in the corresponding micrometer dependency and then using Reporter.fromRegistry to construct a reporter.

For developer convenience we also provide a couple of modules for particular monitoring systems, specifically Datadog and StatsD to provide and ergonomic means for creating reporters for these registries. These can be added to your project as follows:

libraryDependencies += "io.github.keirlawson" %% "meters4s-datadog" % "<see latest version>"

or

libraryDependencies += "io.github.keirlawson" %% "meters4s-statsd" % "<see latest version>"

or

libraryDependencies += "io.github.keirlawson" %% "meters4s-prometheus" % "<see latest version>"

For comprehensive API documentation check the scaladoc.

A simple usage example for incrementing a counter, backed by a Micrometer SimpleMeterRegistry:

import meters4s.{Reporter, MetricsConfig}
import cats.effect.IO

val config = MetricsConfig()
for {
  reporter <- Reporter.createSimple[IO](config)
  counter <- reporter.counter("my.counter")
  _ <- counter.increment
} yield ()

import meters4s.{MetricsConfig, Reporter}
import meters4s.datadog.{DataDog, DataDogConfig}
import cats.effect.IO

val datadog =
  DataDog.createReporter[IO](DataDogConfig(apiKey = "1234"), MetricsConfig())
datadog.use { reporter =>
  reporter.counter("my.counter").flatMap { counter => counter.increment }
}

Here is an example that reports CPU metrics as well as a counter that can be retrieved via the PrometheusMeterRegistry's scrape method. Note your HTTP endpoint will expose this for Prometheus to scrape.

import cats.effect._
import cats.effect.std.Console
import cats.effect.syntax.all._
import cats.syntax.all._
import meter4s.prometheus._
import meters4s.{MetricsConfig, Reporter}
import io.micrometer.core.instrument.binder.system.ProcessorMetrics

import scala.concurrent.duration._

object PromExampleApp extends IOApp {

  override def run(args: List[String]): IO[ExitCode] =
    program[IO](args).as(ExitCode.Success)

  def program[F[_]](args: List[String])(implicit F: Async[F], C: Console[F]): F[Nothing] = {
    val resources =
      for {
        registry <- Prometheus.createMeterRegistry[F](PrometheusConfig())
        _ <- Resource.eval(F.delay(new ProcessorMetrics().bindTo(registry)))
        reporter <- Resource.eval[F, Reporter[F]](
          Prometheus.createReporter[F](MetricsConfig(), registry)
        )
      } yield (registry, reporter)

    resources
      .evalMap { case (registry, reporter) =>
        reporter
          .counter("test_counter")
          .map(counter => (registry, reporter, counter))
      }
      .use { case (registry, reporter, counter) =>
        val scrape = C.println(registry.scrape)
        val increment = counter.increment

        (increment >> scrape).delayBy(10.seconds).foreverM
      }
  }
}

meters4s-http4s implements http4s metrics.

This module records the following meters:

• Timer default.response-time
• Timer default.response-headers-time
• Gauge default.active-requests

The default.response-time timer has the status-code, method and termination tags. The default.response-headers-time timer has the method tag. The default.active-requests does not have any tag.

In addition to these tags, each metric will record the global tags set in the Config.

It is also possible to set a prefix for the metrics name using the prefix configuration setting.

The default name can be customised using a classifier function. With the same classifier function, it is possible to record additional tags using this syntax: classifier[tag1:value1,tag2:value2,tag3:value3]. The classifier part can be blank as well as the tags part can be empty.

The standard tags values are the following:

• statusCode

  • 2xx
  • 3xx
  • 4xx
  • 5xx

• method

  • head
  • get
  • put
  • patch
  • post
  • delete
  • options
  • move
  • trace
  • connect
  • other

• termination

  • normal
  • abnormals
  • error
  • timeout

This library was heavily inspired by (and in some places copied wholesale from) http4s-micrometer-metrics.