Memory utilization of a k8s Pod won't get reduced after executing a task

Hi, I’m using prometheus along with grafana to observe the resource utilization of Pods in the k8s cluster. After getting a request utilization of cpu and memory will be shown in the dashboard. But as expected cpu will be increased and become idle(reduced) gracefully. Even though it’ll take more than expected time to get reduced. (May be couple of minutes after serving the request)

Main problem is memory. I’ve deployed some simple go server (which will do automatic garbage collection), but the memory utilization doesn’t get reduced after executing the relevant task.

But using kubernetes metrics server it shows the that memory and cpu gets reduced as expected after serving a request.