horizontalPodAutoscalers parameter

Use the horizontalPodAutoscalers parameter to define a list of the Horizontal Pod Autoscalers (HPAs) to automatically manage the number of pods in the deployment, based on the observed CPU or custom metric utilization.

  • The horizontalPodAutoscalers enables horizontal pod scaling and assigns it to a deployment. The Operator does not perform scaling. You cannot use horizontalPodAutoscalers with an existing HPA.
  • The horizontalPodAutoscalers must be created by the Operator. The names of the HPAs that are shown in the following YAML are not the actual names for HPAs but logical names that are assigned by the Operator.
  • If the Horizontal Pod Autoscaler (HPA) is defined and the annotation apps.oms.ibm.com/auto-managed is yes, then the Operator gives the replica count control to the HPA and updates the corresponding custom resource instances with the replica in the deployments. For more information about the apps.oms.ibm.com/auto-managed annotation, see Annotations used in Sterling Order Management System Operator .

The following YAML snippet is a sample schema of horizontalPodAutoscalers:

  horizontalPodAutoscalers:
    - name: Hpa1
      maxReplicas: 1
      minReplicas: 1
      metrics:
        - resource:
            name: cpu
            target:
              averageUtilization: 1
              type: Utilization
          type: Resource
    - name: Hpa2
      maxReplicas: 5
      minReplicas: 2
      metrics:
        - resource:
            name: cpu
            target:
              averageUtilization: 5
              type: Utilization
          type: Resource
The following table explains the attributes applicable for the horizontalPodAutoscalers parameter of OMEnvironment.
Table 1. Properties for horizontalPodAutoscalers
Property Default value Value type Required Description
name   string No Specify the name of the Horizontal Pod Autoscaler.
behavior   object No Specify the behavior of the target to scale upwards or downwards by using scaleUp and scaleDown fields. If the behavior is not set, the default HPA scaling rules for scaling up and scaling down are used.
maxReplicas   integer No Specify the maximum limit for the number of replicas to which the HPA can scale up. It cannot be less than minReplicas.
minReplicas   integer No Specify the minimum limit for the number of replicas to which the HPA can scale down. It defaults to 1 pod. Scaling is active until at least one metric value is available.
metrics   array No Specify the metrics details for calculating the replica count. The maximum replica count across all metrics is used. This calculation involves multiplying the ratio between the target value and the current value by the current number of pods. Therefore, the specified metrics decrease as the pod count increases, and vice-versa. For more information about how each specific metric type behaves, see individual metric source types. If the metrics are not set, the system defaults to configuring it to maintain an average CPU utilization of 80%.
For more information about configuring the metrics for the HPA, see Horizontal Pod Autoscaling.