Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Table of Contents

Info
The current work is in a branch. is available since v6.0.0

Enable Metrics

Metrics are disable by default. The gathering of metrics as well as exposing them via HTTP for analysis needs to be enabled first.

Enable HTTP and WebService calls enable the Servlet Filter

Comment in the following block in the openmeetings/WEB-INF/web.xml

Code Block
languagexml
collapsetrue
<!-- Start Prometheus Filter HTTP Servlet metrics
	<filter>
		<filter-name>prometheusFilter</filter-name>
		<filter-class>io.prometheus.client.filter.MetricsFilter</filter-class>
		<init-param>
			<param-name>metric-name</param-name>
			<param-value>webapp_metrics_filter</param-value>
		</init-param>
		<init-param>
			<param-name>help</param-name>
			<param-value>This is the help for your metrics filter</param-value>
		</init-param>
		<init-param>
			<param-name>buckets</param-name>
			<param-value>0.005,0.01,0.025,0.05,0.075,0.1,0.25,0.5,0.75,1,2.5,5,7.5,10</param-value>
		</init-param>
		<init-param>
			<param-name>path-components</param-name>
			<param-value>0</param-value>
		</init-param>
	</filter>

	<filter-mapping>
		<filter-name>prometheusFilter</filter-name>
		<url-pattern>/*</url-pattern>
	</filter-mapping>
	End Prometheus -->

Enable Exposing the metrics and enable Tomcat generic metrics

To enable the export of the metrics as well as some generic Tomcat export (for example around active threads) you need to comment in the following block in openmeetings/WEB-INF/In order to enable the metrics to be published as endpoint comment in the servlet in the web.xml

Code Block
languagexml
collapsetrue
<!-- Start Prometheus export metrics HTTP
	<servlet>
  		<servlet-name>metrics</servlet-name>
  		<servlet-class>io.prometheus.client.exporter.MetricsServlet<class>org.apache.openmeetings.web.util.logging.OpenMeetingsMetricsServlet</servlet-class>
	</servlet>
	<servlet-mapping>
  		<servlet-name>metrics</servlet-name>
  		<url-pattern>/services/metrics/</url-pattern>
	</servlet-mapping>
	End Prometheus -->

This will expose the metrics atSeehttp://$HOST:$PORT/openmeetings/services/metrics/ (be aware of the trailing "/" !)

Warning

If you plan to use this in production you need to secure that endpoint.

For example by:

  • Add an auth security filter in the web.xml so you need credentials to access the endpoint
  • Pointing to a private IP and putting the monitoring into a VPC

Enable metrics for streaming and other application relevant metrics

In order to get insights on the application specific logic and streaming relevant components you need to additionally enable certain annotations by commenting in openmeetingsopenmeetings-web/src/main/webapp/WEB-INF/classes/web.xml#L110applicationContext.xml

Code Block
languagexml
collapsetrue
<!-- Start annotation Prometheus metrics
	<aop:aspectj-autoproxy/>
	End annotation -->

View your metrics in Prometheus

...

Code Block
languageyml
collapsetrue
global:
  scrape_interval:     15s # By default, scrape targets every 15 seconds.
  evaluation_interval: 15s # Evaluate rules every 15 seconds.

  # Attach these extra labels to all timeseries collected by this Prometheus instance.
  external_labels:
    monitor: 'scalyr-blog'

rule_files:
  - 'prometheus.rules.yml'

scrape_configs:
  - job_name: 'prometheus'

    # Override the global default and scrape targets from this job every 5 seconds.
    scrape_interval: 5s

    static_configs:
      - targets: ['localhost:9090']

  - job_name:       'openmeetings-local'

    # Override the global default and scrape targets from this job every 5 seconds.
    scrape_interval: 5s

    metrics_path:  'openmeetings/services/metrics/'

    static_configs:
      - targets: ['192.168.1.66:5080']
        labels:
          group: 'production'

Change 192Change 192.168.1.66:5080 to your local running OpenMeetings instance. Below graphs and queries are done using this docker container.

...

Majority of the metrics collected (as specially the below detailed ones) at are of the Prometheus type "histogram" https://prometheus.io/docs/practices/histograms/

...

Initially you may find it's not quite trivial to calculate the duration for individual calls. But it makes sense once you review below. What we are interested is "rate pre per min" and similar statistics. Not an individual call, but rate/average over a certain period of time.

...

Application metrics based on manual metric

Info

This metric is currently not in use since it can't be disabled

Unfortunate not all places in the code are spring beans. Also sometimes we have inner classes or similar issues.

...