Question

A retail company runs a VPC-native (alias IP) GKE cluster in a custom-mode subnet. The subnet's primary range is used for nodes, and a secondary range /20 is allocated for Pods. After a successful holiday campaign, the platform team plans to triple the number of nodes and Pods over the next year, and they are concerned the current Pod secondary range will be exhausted. They want to add capacity without recreating the cluster or migrating workloads. What should they do?

Accepted Answer

Add a new secondary IP range to the subnet and enable it as an additional Pod range on the cluster using discontiguous multi-Pod CIDR (expandable Pod ranges).

Answer

Delete the existing cluster and recreate it with a larger /16 secondary range for Pods, then redeploy all workloads.

Answer

Convert the cluster to routes-based networking so Pod IPs are drawn dynamically from the VPC route table instead of a fixed secondary range.

Answer

Expand the subnet's primary IP range, since Pods in VPC-native clusters draw their addresses from the primary node range.