Manage dataset configurations

If you complete the product activation and knowledge graph build, but find that the default configuration didn’t provide the desired datasets for your environment, you can edit your dataset configuration. A dataset is a logical grouping of telemetry and can consist of:

Kubernetes metrics
OpenTelemetry metrics, including span or servicegraph metrics
Metrics ingested from AWS that come from Cloud Provider Observability
Prometheus metrics, including rpc or http metrics

During this process, you choose an environment label and, if applicable, a cloud region label. These labels are used to scope your exploration and troubleshooting. For example, you can quickly filter to include only region: us-central-0 and env: prod when troubleshooting a region-specific issue.

Environment label: The environment label corresponds to one or more environments that contain metrics data stored in Grafana Cloud. The environment label populates the list of environments you can select on each knowledge graph page.
If you are using Kubernetes or Prometheus, the environment label is most likely cluster.
However, with OpenTelemetry, the label can be:
- k8s.cluster.name if you run your applications in Kubernetes
- deployment.environment.name
- deployment.environment: You can use this label, but it has been deprecated.
In AWS, the label is account_Id
Environments: By default, the knowledge graph selects All environments, but you can exclude one or more environments.
Cloud region label: The cloud region label populates the list of sites you can select on each knowledge graph page.
Cloud regions: An environment can contain multiple regions. The Cloud regions drop-down defaults to All and you can choose to exclude one or more regions.

If you’ve activated the knowledge graph and don’t see entities being discovered, refer to the troubleshooting guide.

Example metric

In the following example metric, environment="dev" maps to the an environment and cluster="company-abc-infra-dev-euw2-eu-west-2-eks" maps to a site in Asserts.

http_server_duration_bucket{cluster="company-abc-infra-dev-euw2-eu-west-2-eks",environment="dev",,container="support",instance="XXX.XX.XX.XXX:8000",job="support-support-svc-metrics",le="5",namespace="support", service="support-support-svc-metrics"}

However, if you do not have a special environment label, select cluster as your environment label.

Before you begin

Before you edit a dataset configuration, ensure the following:

You are sending Kubernetes or OpenTelemetry metrics to Grafana Cloud
You are familiar with your metrics coming into Grafana Cloud, including the source environment and clusters
You have Grafana Cloud Admin permissions

Refer to one of the following dataset types: