Graphite Integration

Build Status

Write report metrics in graphite format.

Usage

Since kamon 2.0 you only have to include the dependency to the reporter.

libraryDependencies += "io.kamon" %% "kamon-graphite" % "2.0.0"

Example config

application.conf

kamon {
  environment {
    service = "supercool-app"
    tags {
      env = "local"
    }
  }
  graphite {
    hostname = "127.0.0.1"
    port = 2003
  }
}      

Config

Per default the graphite tag support (available since graphite 1.1) is enabled (see include-tags feature flag). These are the relevant config values that chan be adapted to your needs using a custom application.conf.

kamon {
    modules {
      graphite-reporter {
        enabled = true
      }
    }

    graphite {
        # Hostname and port in which your Carbon daemon is running.
        hostname = "127.0.0.1"
        port = 2003

        # Prefix for all metrics sent to Graphite.
        metric-name-prefix = "kamon-graphite"

        # instead of adding tags as suffix to metric name (format graphite 1.1) metricname;tag1=value1;tag2=value2
        # the metric will be named metricname.tag1.value1.tag2.value2
        # this is useful for older graphite versions (prior to 1.1) without tagging support
        legacy-support = false

        # For histograms, which percentiles to count
        percentiles = [50.0,90.0,99.0]

        # Allow including environment information as tags on all reported metrics.
        environment-tags {

          # Define whether specific environment settings will be included as tags in all exposed metrics. When enabled,
          # the service, host and instance tags will be added using the values from Kamon.environment().
          include-service = yes
          include-host = yes
          include-instance = yes

          # Specifies which Kamon environment tags should be ignored. All unmatched tags will be always added to all metrics.
          exclude = []
        }

        tag-filter {
          includes = ["**"]
          excludes = []
        }
    }
}

Failure handling

When sending metrics at a tick interval fails the current snapshot will be dropped and the next snapshot will try to send metrics using new connection.