HorizontalPodAutoscaler Memory CPU calculation details explored in OpenShift 4

Issue

Illustrated below is an instance showcasing the outcome when utilizing HPA API version 2, employing memory and CPU metrics as benchmarks for automatic scaling:

$ oc get hpa -n nginx
NAMESPACE   NAME            REFERENCE                    TARGETS             MINPODS   MAXPODS   REPLICAS   AGE
nginx       nginx-ingress   Deployment/nginx-ingress     4%/50%, 6%/50%    2         10        2          198d

Let's begin by comprehensively grasping the output delivered by the aforementioned command. Within the "nginx" namespace, there exists a deployment labeled as "nginx-ingress," managed by an HPA named "nginx-ingress."

The HPA is employed in this context to dynamically expand or contract the specified deployment, ensuring a minimum of 2 pods and a maximum of 10. Presently, this deployment houses 2 pods (termed replicas), executing the responsibilities of an NGINX Ingress Controller.

In the configuration of this cluster, the term "targets" denotes the utilization of memory metrics (4%/50%). Here, the prevailing computed average memory usage registers at 4%, while the designated scaling threshold is set at 50%. Similarly, the second metric pertains to CPU (6%/50%), applying an identical rationale.