Grafana Cloud

Configuration deployment strategies

Configuration in Grafana Fleet Management is a powerful and flexible tool for managing collectors at scale. Learn how to optimize your configuration deployments for your unique observability setup.

Remote configuration only

If you want to configure your collectors fully remotely, you can move the contents of local configuration files to configuration pipelines in Fleet Management and add the minimum required configuration, such as the remotecfg block for Alloy, to the local configuration files.

This method gives you maximum control over which configuration pipelines are in use. Turn pipelines off with a click of the switch. Combine them using matching attributes to build complete remote configurations based on each collector’s characteristics and your observability needs.

Remote configuration is highly customizable, even at scale. Use it to optimize your workflows. For example:

  • Universal pipelines: Create a single configuration pipeline in Fleet Management and assign it to every collector with a matching attribute. As new collectors are registered with Fleet Management, you don’t need to modify each local configuration file or reload the collector. Instead, add the attribute to new collectors as you register them, and they automatically receive the pipeline when they poll the Fleet Management server for remote configuration. To see an example, refer to Apply the MySQL integration to every collector.
  • Dynamic pipelines: If you need to enable or disable a pipeline on demand, place it in remote configuration. Remote configuration pipelines can be activated or deactivated with a switch, without requiring changes to multiple local files. For specialized monitoring use cases such as profiling, create a separate pipeline in remote configuration with matching attributes that target a specific subset of collectors. You can selectively enable resource-intensive monitoring only when and where you need it. To see an example, refer to Collect on-demand profiles to diagnose excessive resource usage.

To learn how to create remote configuration pipelines, refer to Create and assign a configuration pipeline.

Hybrid configuration

You can continue configuring some parts of your collectors locally and create supplemental configuration pipelines in Fleet Management. Alloy collectors run their local and remote configurations in parallel. Because of this isolation, local and remote Alloy configurations cannot share components between them. They must be independently complete so that each one includes all of its own referenced components.

OpenTelemetry Collectors managed by the OpAMP Supervisor merge their local and remote configurations.

Local configuration only

Although configuration pipelines are a central feature of Fleet Management, they are optional. You can choose to configure your collectors locally and use Fleet Management only to monitor their status.

Disable remote configuration

Turn off autogenerated self-monitoring pipelines to disable remote configuration.

Caution

Disabling the autogenerated pipelines limits your ability to monitor your collector fleet. Without these pipelines, collectors don’t send their internal telemetry to Fleet Management, and the application interface can’t display any data in the health dashboards and logs panel of the collector details view. If you want to see health monitoring data in Fleet Management, you must adjust your local configuration as described in the following task.

  1. Register your collectors with the Fleet Management service.

  2. In your Grafana Cloud stack, click Connections > Fleet Management.

  3. Confirm that your registered collectors appear in the Inventory tab.

  4. Switch to the Remote configuration tab.

  5. Disable the autogenerated self-monitoring configuration pipelines by turning the Active switch off.

  6. To see health and status information in Fleet Management, update your local configuration as follows:

    For Alloy collectors:

    • Add the collector_id label to your local configuration. The value of the collector_id label must match the id set in the remotecfg block.
      • Add collector_id as an external label on all remote_write components, or
      • Create more targeted configuration pipelines to collect internal logs and metrics, adding the collector_id label only to those pipelines.

    For OpenTelemetry Collectors:

    • Add the service::telemetry configuration for internal metrics, logs, and traces to each local configuration and export the telemetry to Grafana Cloud. You can copy a pre-filled version of the self-monitoring pipeline from the Remote configuration tab in Fleet Management. Look for the self_monitoring_otelcol configuration pipeline in the list, copy the contents, and paste them into the Collector’s local configuration file. If your Collector is deployed with Kubernetes, also make sure to copy the contents of the self_monitoring_otelcol_kube_stack pipeline and add it to the local config. Make sure you’ve set all necessary environment variables.