S3 CSI Driver v2 Causes Increased Pod and IP Consumption: Mitigating Scaling Issues with Intermediate Mountpoint Pods

Introduction & Problem Statement The integration of cloud-native storage solutions, particularly the S3 CSI driver v2 , has exposed a critical scalability challenge in Kubernetes environments. Central to this issue is the driver's reliance on intermediate Mountpoint pods , which serve as a proxy layer for mounting S3 directories into application pods. While this architecture enhances storage accessibility, it introduces a resource consumption model that scales linearly with the number of mounts, posing a significant threat to system scalability. Mechanism of Resource Exhaustion In a cluster hosting 350 deployments , the S3 CSI driver v2 instantiates one Mountpoint pod per mount , resulting in approximately 450 additional pods within the mount-s3 namespace. Each Mountpoint pod is assigned a unique VPC IP address , a constrained resource in cloud infrastructures. The resource exhaustion process unfolds as follows: Resource Allocation: Kubernetes schedules each Mountpoint pod as an indepe