You're viewing version 2.12 of the OpenSearch documentation. This version is no longer maintained. For the latest version, see the current documentation. For information about OpenSearch version maintenance, see Release Schedule and Maintenance Policy.

This is an experimental feature and is not recommended for use in a production environment. For updates on the progress of the feature or if you want to leave feedback, see the associated GitHub issue.

While the OpenSearch Stats APIs offer insight into the inner workings of each node and an OpenSearch cluster as a whole, the statistics lack certain details, such as percentiles, and do not provide the semantics of richer metric types, like histograms. Consequently, identifying outliers within cluster statistics becomes challenging when using only the Stats API.

The metrics framework feature adds comprehensive metrics support to effectively monitor an OpenSearch cluster. Using the Metrics Framework APIs, plugin, and extension, developers can add new monitoring metrics. In addition, the OpenSearch distribution bundles the telemetry-otel plugin, which provides the implementation for metrics instrumentation based on the OpenTelemetry Java SDK.

Getting started

The metrics framework feature is experimental as of OpenSearch 2.11. To begin using the metrics framework feature, you need to first enable the telemetry feature by using the opensearch.experimental.feature.telemetry.enabled feature flag and subsequently by using the metrics framework feature flag.

Enabling this feature can consume system resources. Before enabling the metrics framework feature, determine whether you have sufficient cluster resources to allocate.

Enabling the feature flag on a node using tarball

The enable flag is toggled using a Java Virtual Machine (JVM) parameter that is set either in OPENSEARCH_JAVA_OPTS or in config/jvm.options.

Option 1: Enable the experimental feature flag in the opensearch.yml file

1. Navigate to your OpenSearch directory using the following command:

  cd \path\to\opensearch

1. Open your opensearch.yaml file.
2. Add the following setting to opensearch.yaml:

  opensearch.experimental.feature.telemetry.enabled=true

1. Save your changes and close the file.

Option 2: Modify jvm.options

To enable the metrics framework feature using jvm, add the following line to config/jvm.options before starting OpenSearch:

-Dopensearch.experimental.feature.telemetry.enabled=true

Option 3: Enable from an environment variable

You can enable the metrics framework feature with a single command by adding the metrics framework environment variable to the OPENSEARCH_JAVA_OPTS command, as shown in the following example:

OPENSEARCH_JAVA_OPTS="-Dopensearch.experimental.feature.telemetry.enabled=true" ./opensearch-2.9.0/bin/opensearch

You can also define the environment variable separately before running OpenSearch by running the following command:

export OPENSEARCH_JAVA_OPTS="-Dopensearch.experimental.feature.telemetry.enabled=true"
 ./bin/opensearch

Enable with Docker

If you’re running OpenSearch using Docker, add the following line to docker-compose.yml under environment:

OPENSEARCH_JAVA_OPTS="-Dopensearch.experimental.feature.telemetry.enabled=true"

Enable the metrics framework feature

Once you’ve enabled the feature flag, you can enable the metrics framework feature by using the following setting, which enables metrics in the opensearch.yaml file:

telemetry.feature.metrics.enabled=true

The metrics framework feature supports various telemetry solutions through plugins. Use the following instructions to enable the telemetry-otel plugin:

1. Publish interval: The metrics framework feature can locally aggregate metrics with unique information about the configured publish interval and then export those metrics. By default, the interval is 1 minute. However, you can change the interval using the telemetry.otel.metrics.publish.interval cluster setting.
2. Exporters: Exporters are responsible for persisting the data. OpenTelemetry provides several out-of-the-box exporters. OpenSearch supports the following exporters:
   • LoggingMetricExporter: Exports metrics to a log file, generating a separate file in the logs directory _otel_metrics.log. Default is telemetry.otel.metrics.exporter.class=io.opentelemetry.exporter.logging.LoggingMetricExporter.
   • OtlpGrpcMetricExporter: Exports spans through gRPC. To use this exporter, you need to install the otel-collector on the node. By default, it writes to the http://localhost:4317/ endpoint. To use this exporter, set the following static setting: telemetry.otel.metrics.exporter.class=io.opentelemetry.exporter.otlp.metrics.OtlpGrpcMetricExporter.