You oversee a retail website composed of multiple microservices deployed in a GKE Standard node pool with node autoscaling enabled. Each microservice has defined resource limits and a Horizontal Pod Autoscaler (HPA) configured. During peak traffic, alerts trigger for one microservice. Upon investigation, half of its Pods show an OOMKilled status, and the Pod count remains at the minimum HPA limit. How should you address this issue?