eclair/docs/Monitoring.md

Monitoring Eclair

Eclair uses Kamon to generate metrics and spans. Kamon automatically collects many akka-related metrics and performance counters from the underlying system.

Kamon uses an agent (Kanela) that runs alongside eclair and periodically sends metrics and spans to a monitoring backend.

Enabling monitoring with Kamon APM

Monitoring is disabled by default. To enable monitoring with Kamon's hosted platform, create an account on their website and add the following to your eclair.conf:

eclair.enable-kamon=true

kamon {

  apm {
    # Put the API key obtained from your Kamon account.
    api-key=XXXXXXX
  }

  trace {
    # Configures a sampler that decides which traces should be reported to the trace backends. The possible values are:
    #   - always: report all traces (will impact application performance).
    #   - never:  don't report any trace.
    #   - random: randomly decide using the probability defined in the random-sampler.probability setting.
    #   - adaptive: keeps dynamic samplers for each operation while trying to achieve a set throughput goal.
    sampler = "random"
  }

}

When starting eclair, you should enable the Kanela agent:

eclair.sh -with-kanela

Your eclair node will start exporting monitoring data to Kamon. You can then start creating dashboards, graphs and alerts directly on Kamon's website.

Enabling monitoring with Prometheus

Kamon supports many other monitoring backends. This can be useful for nodes that don't want to export any data to third-party servers.

Eclair currently supports exporting metrics to Prometheus. To enable monitoring with Prometheus, add the following to your eclair.conf:

eclair.enable-kamon=true

// The Kamon APM reporter is enabled by default, but it won't work with Prometheus, so we disable it.
kamon.modules {
  apm-reporter {
    enabled = false
  }
}

kamon {
  prometheus {
    start-embedded-http-server = yes
    embedded-server {
      hostname = 0.0.0.0
      port = <port to expose to prometheus>
    }
  }
}

You should then configure your Prometheus process to scrape metrics from the exposed http server.

• Download Prometheus here.
• Add the following configuration to the prometheus.yml file (see the Prometheus documentation for more details)

global:
  scrape_interval:     15s # By default, scrape targets every 15 seconds.

scrape_configs:
  - job_name: 'eclair'
    static_configs:
      - targets: ['<url of the eclair http embedded server>']

Example metrics

Apart from akka and system metrics, eclair generates a lot of lightning metrics. The following metrics are just a small sample of all the metrics we provide:

• Number of local channels, grouped by their current state (offline, normal, closing, etc)
• Relayed payments and fees collected
• Number of connected peers
• Bitcoin wallet balance
• Various metrics about the public graph (nodes, channels, updates, etc)