Configure container resources and replicas

This pages describes how to configure container resources and replicas for controllers in Config Connector.

Configure resource allocation for a Config Connector controller container

In Config Connector version 1.106 and later, you can configure the CPU and memory (RAM) resources allocated to a container in a Config Connector controller Pod. You can configure the following controllers:

• cnrm-webhook-manager
• cnrm-controller-manager
• cnrm-deletiondefender
• cnrm-resource-stats-recorder
• cnrm-unmanaged-detector (Config Connector version 1.108 and later)

There are two types of custom resources used for configuration: ControllerResource and NamespacedControllerResource.

Use ControllerResource for cluster-scoped components

Use the ControllerResource custom resource to configure resources for Config Connector components that run as cluster-scoped deployments.

The following components are always cluster-scoped, even when Config Connector is installed in namespaced mode:

• cnrm-webhook-manager
• cnrm-deletiondefender
• cnrm-resource-stats-recorder
• cnrm-unmanaged-detector

The cnrm-controller-manager is also cluster-scoped when Config Connector is installed in cluster mode.

For example, to configure the memory request and limit of the webhook container of the cnrm-webhook-manager controller, create and apply the following YAML file:

apiVersion: customize.core.cnrm.cloud.google.com/v1beta1
kind: ControllerResource
metadata:
  name: cnrm-webhook-manager
spec:
  containers:
    - name: webhook
      resources:
        limits:
          memory: 512Mi
        requests:
          memory: 256Mi

Use NamespacedControllerResource for the namespaced controller manager

Use the NamespacedControllerResource custom resource only to configure resources for the cnrm-controller-manager when Config Connector is running in namespaced mode.

The configuration of container resources for a namespaced controller is enabled in Config Connector version 1.108 and later.

The following YAML file shows an example configuration:

apiVersion: customize.core.cnrm.cloud.google.com/v1beta1
kind: NamespacedControllerResource
metadata:
  name: cnrm-controller-manager # name should not contain the namespace ID suffix
  namespace: NAMESPACE
spec:
  containers:
    - name: manager
      resources:
        limits:
          cpu: 200m
          memory: 512Mi
        requests:
          cpu: 100m
          memory: 256Mi

Replace NAMESPACE with the name of your namespace.

Configure replicas for a Config Connector controller Pod

In Config Connector version 1.107 and later, you can configure the number of replicas for a Config Connector controller. You can only configure the cnrm-webhook-manager.

For example, the following steps show how to configure the number of replicas for the cnrm-webhook-manager controller to 3.

1. Create a file named configure-webhook-manager.yaml and copy the following YAML into it:

   apiVersion: customize.core.cnrm.cloud.google.com/v1beta1
   kind: ControllerResource
   metadata:
     name: cnrm-webhook-manager
   spec:
     replicas: 3
     containers:
       - name: webhook
         resources:
           limits:
             memory: 512Mi
           requests:
             memory: 256Mi
   

2. Use kubectl apply to apply the modified configuration to your cluster:

   kubectl apply -f configure-webhook-manager.yaml

3. Verify the successful configuration by running the following command.

   kubectl get controllerresource cnrm-webhook-manager -o jsonpath='{.status.healthy}'

   It should display status.healthy field set to true.

4. Verify the presence of 3 webhook Pods within your cluster.

   kubectl get pods -n cnrm-system -l cnrm.cloud.google.com/component=cnrm-webhook-manager

   Creating and recreating the Pods could take a few minutes.

Note: Config Connector includes a HorizontalPodAutoscaler that periodically adjusts the appropriate replicas for the cnrm-webhook-manager controller. If you manually configure the replicas using the configuration YAML, the Config Connector operator automatically adjusts the minReplicas and maxReplicas of the HorizontalPodAutoscaler to ensure the cluster maintains at least the specified number of replicas in the configuration YAML.

Revert configuration changes in container resources and replicas

Use kubectl delete to delete the ControllerResource custom resources that you configured. After deletion, the Config Connector operator recreates the controllers with the default container resources and replicas.

kubectl delete -f configure-webhook-manager.yaml