A Helm chart to deploy the New Relic Prometheus OpenMetrics integration

Homepage: https://docs.newrelic.com/docs/infrastructure/prometheus-integrations/install-configure-openmetrics/configure-prometheus-openmetrics-integrations/

You can install this chart using nri-bundle located in the helm-charts repository or directly from this repository by adding this Helm repository:

helm repo add nri-prometheus https://newrelic.github.io/nri-prometheus
helm upgrade --install newrelic-prometheus nri-prometheus/nri-prometheus -f your-custom-values.yaml

• https://github.com/newrelic/nri-prometheus
• https://github.com/newrelic/nri-prometheus/tree/main/charts/nri-prometheus

When a service is labeled or annotated with scrape_enabled_label (defaults to prometheus.io/scrape), nri-prometheus will attempt to hit the service directly, rather than the endpoints behind it.

This is the default behavior for compatibility reasons, but is known to cause issues if more than one endpoint is behind the service, as metric queries will be load-balanced as well leading to inaccurate histograms.

In order to change this behaviour set scrape_endpoints to true and scrape_services to false. This will instruct nri-prometheus to scrape the underlying endpoints, as Prometheus server does.

Existing users that are switching to this behavior should note that, depending on the number of endpoints behind the services in the cluster the load and the metrics reported by those, data ingestion might see an increase when flipping this option. Resource requirements might also be impacted, again depending on the number of new targets.

While it is technically possible to set both scrape_services and scrape_endpoints to true, we do no recommend doing so as it will lead to redundant metrics being processed,

This chart implements the New Relic's common Helm library which means that it honors a wide range of defaults and globals common to most New Relic Helm charts.

Options that can be defined globally include affinity, nodeSelector, tolerations, proxy and others. The full list can be found at user's guide of the common library.

See this snippet from the values.yaml file:

global:
  lowDataMode: false
lowDataMode: false

To reduce the amount ot metrics we send to New Relic, enabling the lowDataMode will add these transformations:

transformations:
  - description: "Low data mode defaults"
    ignore_metrics:
      # Ignore the following metrics.
      # These metrics are already collected by the New Relic Kubernetes Integration.
      - prefixes:
        - kube_
        - container_
        - machine_
        - cadvisor_

• alvarocabanas
• carlossscastro
• sigilioso
• gsanchezgavier
• kang-makes
• marcsanmi
• paologallinaharbur
• roobre