Collecting Metrics from Prometheus Endpoints

A blog post by John Mazzitelli

agent | hawkular | prometheus

Hawkular WildFly Agent 0.17.2.Final has just been released. This new release introduces a very interesting feature.

Prometheus Metrics

The agent now has the ability to monitor Prometheus endpoints and store their metrics to Hawkular Metrics. This means any component that exports Prometheus metric data via either the Prometheus binary or text formats can have those metric data collected by the agent and pushed up into Hawkular Metrics.


It is very easy to configure the Hawkular WildFly Agent to collect Prometheus metric data. In the WildFly container that houses the agent, edit its standalone.xml configuration file and add one or more <remote-prometheus> elements in the <managed-servers> section of the agent subsystem:

<remote-prometheus name="My Remote Prometheus Endpoint"
                   timeUnits="seconds" />

This tells the agent to periodically scrape the metric data from that remote Prometheus endpoint and whatever metrics it finds will be pushed to Hawkular Metrics.

In fact, if you only want to collect Prometheus metric data (and not collect any other kind of metric data, like those from DMR endpoints or Jolokia/JMX endpoints) you need only those <remote-prometheus> declarations. You do not have to also collect from <remote-dmr> or <remote-jmx> servers (though you are free to do so if you want to monitor types of those components, too).

Some Prometheus endpoints expose a lot of metrics, some of which you may not be interested in collecting and storing. In this case, you have the option to define metric sets which tell the agent which metrics to collect and store (any others will be ignored).

Here is an example of a full agent subsystem configuration that defines a remote Prometheus endpoint and the metrics to be collected from that endpoint:

<subsystem xmlns="urn:org.hawkular.agent:agent:1.0" enabled="true">
    <diagnostics enabled="true" interval="1" reportTo="LOG" timeUnits="minutes"/>
    <storage-adapter type="HAWKULAR" serverOutboundSocketBindingRef="hawkular" username="jdoe" password="password"/>

    <metric-set-prometheus name="Important Metrics">
        <metric-prometheus name="http_.*"/>
        <metric-prometheus name="evaluation_failures_total" />

        <remote-prometheus enabled="true"
                           name="My Remote Prometheus Endpoint"
                           metricSets="Important Metrics"
                           timeUnits="seconds" />

The above configuration tells the agent to scrape the given Prometheus endpoint URL every 30 seconds and store all the metrics with the name "evaluation_failures_total" or with a name that matches the regular expression "http_.*". Any other metrics exported by that remote Prometheus endpoint will be ignored.

Information about this release can be found on Github here:

Published by John Mazzitelli on 22 April 2016


© 2016 | Hawkular is released under Apache License v2.0