Namespace Actions

ScaleOps supports defining policies and automating a namespace using CRD or annotations.

Annotations

You can use annotations to automate a namespace and set default policies for all its workloads.

Automating a namespace using an annotation

Control namespace’s automation with annotation.

Use scaleops.sh/default-auto: "true" to set default automation for Workload Rightsizing. To disable automation, delete the annotation or set its value to “false”.

For GPU Rightsizing, use scaleops.sh/default-gpu-auto: "true" to enable GPU automation for all workloads in the namespace.

Example:

apiVersion: v1
kind: Namespace
metadata:
  annotations:
    scaleops.sh/default-auto: "true"

Setting namespace policy

Control namespace’s default Workload Rightsizing policy with annotation.

Use scaleops.sh/default-policy: {policy name} to set a specific policy for your namespace.

For GPU Rightsizing, use scaleops.sh/default-gpu-policy: {policy name} to set a default GPU policy for all workloads in the namespace.

Example:

apiVersion: v1
kind: Namespace
metadata:
  annotations:
    scaleops.sh/default-policy: production
    scaleops.sh/default-gpu-policy: production

Excluding namespace

Exclude namespace from Workload Rightsizing automation by adding scaleops.sh/exclude-automation: "true" annotation.

ScaleOps will not allow automation on this excluded namespaces. If the namespace has automated workloads, adding the annotation will disable their automation.

Removing this annotation after excluding a namespace will allow the recommendations to be controlled again and will be automatically automated if the namespace or cluster is automated.

Example:

apiVersion: v1
kind: Namespace
metadata:
  annotations:
    scaleops.sh/exclude-automation: "true"

Custom namespace labels

Available in v1.26.10+

ScaleOps enhances namespace management by enabling you to include or exclude namespaces from automation based on namespace labels. This allows workloads to remain visible for troubleshooting while controlling their automation status.

Configuration can be done via:

• Helm chart: Use the namespaceAutomation section in your values file. See Helm configuration reference.
• Console UI: Navigate to Settings → General to select labels.
• ConfigMap: Create the scaleops-custom-namespace-labels ConfigMap manually in the scaleops-system namespace.

Configuration Options

ConfigMap Example

apiVersion: v1
kind: ConfigMap
metadata:
  name: scaleops-custom-namespace-labels
  namespace: scaleops-system
data:
  exclude-labels: '["env=staging", "team=test-.*"]'
  include-labels: '["scaleops.sh/auto=true", "env=production"]'

Helm Values Example

namespaceAutomation:
  excludeLabels:
    - "env=staging"
    - "team=test-.*"  # regex pattern for value
  includeLabels:
    - "scaleops.sh/auto=true"
    - "env=production"

Ignored Namespaces

Ignored namespaces are completely hidden from ScaleOps — workloads inside them are not displayed in the dashboard and not optimized in any way.

Configuration can be done via:

• Console UI: Navigate to Settings → General under Namespace operations → Ignored Namespaces.
• Helm chart: Use the ignoredNamespaces field in your values file. It accepts a list of namespace names or regex patterns. See Helm configuration reference.

Default namespaces:

ignoredNamespaces:
  - kube-system
  - kube-public
  - kube-node-lease
  - "openshift.*"
  - trident
  - cluster-autoscaler

Example — adding custom ignored namespaces:

ignoredNamespaces:
  - kube-system
  - kube-public
  - kube-node-lease
  - "openshift.*"
  - trident
  - cluster-autoscaler
  - my-internal-namespace
  - "temp-.*"

configParams:
  overrideUserKubeSystemIgnoredNamespace: "true"

Note: A few ScaleOps features still operate on ignored namespaces: custom workloads receive their ScaleOps label regardless, and Pod Placement and Spot Optimization continue to function on kube-system namespace.

CRD

The AutomatedNamespace CRD allows you to automate an entire namespace or set a policy for all workloads within that namespace across all optimization products.

To apply this, create an AutomatedNamespace CR in the specific namespace you wish to automate or set a policy for.

For granular control over each optimization product, use the following product-specific fields:

• Rightsizing: rightsizeOptimize: true/false, defaultRightsizePolicy: "policy-name"
• Replicas: replicasOptimize: true/false, defaultReplicasPolicy: "policy-name"
• Spot: spotOptimize: true/false, defaultSpotPolicy: "policy-name"
• GPU Rightsizing: gpuOptimize: true/false, defaultGpuPolicy: "policy-name"
• Java Optimization: javaOptimize: true/false

By adding excludeAutomation field, you can exclude the entire namespace from automation.

Example - Configure namespace with ANS CRD:

apiVersion: analysis.scaleops.sh/v1alpha1
kind: AutomatedNamespace
metadata:
  name: name-of-namespace # needs to match the name of the namespace
  namespace: name-of-namespace # needs to match the name of the namespace
spec:
  rightsizeOptimize: true # enable or disable rightsize automation for all workloads in the namespace
  defaultRightsizePolicy: "production"
  replicasOptimize: true # enable or disable replicas automation for all workloads in the namespace
  defaultReplicasPolicy: "cost"
  spotOptimize: true # enable or disable spot automation for all workloads in the namespace
  defaultSpotPolicy: "spot-friendly"
  gpuOptimize: true # enable or disable GPU rightsize automation for all workloads in the namespace
  defaultGpuPolicy: "production"
  javaOptimize: true # enable or disable Java optimization automation for all workloads in the namespace
  excludeAutomation: false 

Note: The optimize field is supported to enable or disable automation for Workload Rightsizing for backward compatibility. The defaultPolicy field is supported to apply a specific policy to all workloads for Workload Rightsizing for backward compatibility.