Manage existing instrumentation with Agent Control

If you would like to try out Agent Control on a cluster that is already instrumented with New Relic, you will need to remove the existing New Relic agents and manage them through the Agent Control. All features and all configuration exposed by the Helm chart of each agent can still be applied in a unified way through the Agent Control.

Match the existing instrumentation with the new agents managed by the Agent Control

First of all, match your existing instrumentation with the supported agents.

You can deploy and manage through the Agent Control:

• The following Helm charts included in the nri-bundle chart:
  • New Relic-infrastructure, nri-kube-events, kube-state-metrics, and nri-kube-events (now managed through the New Relic Infrastructure agent type)
  • New Relic-prometheus-configurator (now managed through the New Relic Prometheus agent type)
  • New Relic-logging (now managed through the Fluent Bit agent type)
• The nr-k8s-otel-collector chart (now managed through the New Relic OpenTelemetry Collector agent type)

If you installed agents using Helm, retrieve your installation config from the cluster by running the following command:

$
$ helm get values <release-name> --namespace <namespace-name>

If you have the file saved from the installation, you don't have to do anything.

On the other hand, if you have instrumented the cluster via manifest, find the corresponding options in each agent chart.

Create the values.yaml config to install the Agent Control

Start the Agent Control Installation via guided install and select all the agents you are interested in. Copy the values.yaml file to your local machine. Before applying it, update each agent configuration to correspond with the configurations of the agents that are already running in the cluster.

Uninstall the agents you want to control

Uninstall the agents and components that you want to manage via Agent Control. Refer to the existing agent documentation for instructions on how to uninstall them.

Helm installation

After you have modified your values.yaml file, continue with the installation steps included in the guided install.

Example

We had a cluster already instrumented via Helm charts. We have retrieved the following values.yaml from a release of nri-bunle:

global:
  cluster: test-migration
  licenseKey: ***
kube-state-metrics:
  enabled: false
New Relic-prometheus-agent:
  enabled: true
New Relic-infrastructure:
  enabled: true
  kubelet:
    tolerations:
      - operator: "Exists"
        effect: "NoSchedule"
      - operator: "Exists"
        effect: "NoExecute"
      - operator: "Exists"
        key: "MyToleration"
  ksm:
    enabled: false
  common:
    config:
      interval: 29s
New Relic-logging:
  enabled: true
  image:
    tag: "latest"
  resources:
    limits:
      cpu: 200m
    requests:
      cpu: 200m
nri-kube-events:
  enabled: true
  customAttributes: 
    test_tag_label: test_tag_value

Notice that nri-kube-events, New Relic-logging and New Relic-infrastructure have a configuration that needs to be reused for setup with Agent Control. On the other hand, New Relic-prometheus-agent was installed as part of nri-bundle and had no additional configuration. Moreover, we had the KSM component set to false.

We then retrieve the configuration created via the guided install and modify the values.yaml file so that the agents you plan to manage are enabled while maintaining the same settings as the uninstalled agents.

global:
  cluster: "test-migration"
  licenseKey: "****"
agent-control-deployment:
  identityClientId: "****"
  identityClientSecret: "****"
  config:
    fleet_control:
      fleet_id: "****"
      auth:
        organizationId: "****"
    subAgents:
      logs:
        type: New Relic/io.fluentbit:0.1.0
        content:
          chart_version: "1.25.1"
          chart_values:
            New Relic-logging:
                image:
                  tag: "latest"
                resources:
                  limits:
                    cpu: 200m
                  requests:
                    cpu: 200m
      infrastructure:
        type: New Relic/com.New Relic.infrastructure:0.1.0
        content:
          chart_version: "5.0.109"
          chart_values:
            New Relic-infrastructure:
              kubelet:
                tolerations:
                  - operator: "Exists"
                    effect: "NoSchedule"
                  - operator: "Exists"
                    effect: "NoExecute"
                  - operator: "Exists"
                    key: "MyToleration"
              ksm:
                enabled: false
              common:
                config:
                  interval: 29s
            nri-kube-events:
              customAttributes: 
                test_tag_label: test_tag_value
            kube-state-metrics:
              enabled: false
      prometheus:
        type: New Relic/com.New Relic.prometheus:0.1.0
        content:
          chart_version: "1.15.4"

Next, we delete the old instrumentation:

$
helm delete my-installation -n New Relic

Finally, we create the new one via Agent Control:

$
helm upgrade --install agent-control -n New Relic New Relic/agent-control --create-namespace --values my_migrated_values.yaml