Cortex Time Series Plugin

The Cortex Time Series plugin enables Meridian to persist time series performance and latency data to Cortex. The plugin uses the time series storage integration layer to convert metrics to a Prometheus model and delegate writes and reads to Cortex. It also lets you visualize graphs from the Cortex instance in the OpenNMS plugin for Grafana through the OpenNMS performance data source or the Prometheus data source.

Architecture diagram displaying the Cortex Time Series plugin and how it integrates with Meridian
Figure 1. Cortex Time Series plugin architecture


You must have the following before you start to use the Cortex plugin:

  • Horizon 30.0.0+ (Meridian availability with the 2023 release)

  • Cortex version 1.10.0+

We plan to keep the plugin compatible with Mimir, Grafana’s fork of Cortex, as long as Mimir remains compatible with Cortex. The minimum supported Mimir version is 2.0.0.


To use the Cortex Time Series plugin, follow these steps:

  1. Start Cortex (see the Cortex documentation for instructions). If your aim is to get up to speed quickly in a test environment, you can use Docker Compose to deploy a minimal, standalone Cortex node.

  2. Download the plugin’s .kar file from GitHub.

  3. Copy the .kar file into ${OPENNMS_HOME}/deploy.

  4. Enable and configure time series database plugins in ${OPENNMS_HOME}/etc/

    org.opennms.timeseries.strategy=integration (1) (2)
    org.opennms.timeseries.tin.metatags.tag.node=${node:label} (3)
    1 Make sure that the org.opennms.timeseries.strategy property is set only once so you don’t have conflicting values in multiple properties files.
    2 If you are enabling the dual write plugin on an existing installation and want to keep historical metrics, do not set the org.opennms.timeseries.strategy to integration until you have written enough data to expire any collected data prior to enabling the dual write feature.
    3 See the time series integration section for more information on defining metatags.
    If you are deploying to a container, instead of setting org.opennms.timeseries.strategy via a properties file, you can set your container’s environment variables configuration to specify OPENNMS_TIMESERIES_STRATEGY=integration.
  5. Configure the Cortex Time Series plugin in Meridian via the Karaf shell, or by creating or modifying ${OPENNMS_HOME}/etc/org.opennms.plugins.tss.cortex.cfg.

    The writeUrl and readUrl properties are required. All other properties shown here are example values to overwrite the defaults. Monitor your Cortex cluster’s performance and adjust these properties as necessary. Substitute the correct host and port for your Cortex cluster.

    • Karaf shell

    • CFG file

    Configure the plugin’s properties:
    config:edit org.opennms.plugins.tss.cortex
    property-set writeUrl http://cortex-service-host:9009/api/prom/push
    property-set readUrl http://cortex-service-host:9009/prometheus/api/v1
    property-set maxConcurrentHttpConnections 100
    property-set writeTimeoutInMs 1000
    property-set readTimeoutInMs 1000
    property-set metricCacheSize 1000
    property-set externalTagsCacheSize 1000
    property-set bulkheadMaxWaitDurationInMs 9223372036854775807
    Create or modify the ${OPENNMS_HOME}/etc/org.opennms.plugins.tss.cortex.cfg file:
  6. Edit or create ${OPENNMS_HOME}/etc/featuresBoot.d/cortex.boot to configure the plugin to remain installed after Meridian restarts:

    opennms-plugins-cortex-tss wait-for-kar=opennms-cortex-tss-plugin
  7. Restart the Meridian service.

Cortex tips

Navigate to the following URLs to view information about your Cortex server:

  • Server status: http://cortex-service-host:9009

  • Ring: http://cortex-service-host:9009/ring

  • Internal metrics: http://cortex-service-host:9009/metrics