MENU
You have the following routing design. You discover that Compute Engine instances in Subnet-2 in the asia-southeast1 region cannot communicate with compute resources on-premises. What should you do?
It allows you to use any public IP addresses that are not owned by Google or your organization for your Pods, which can help mitigate address exhaustion in your enterprise.
It prevents any external traffic from reaching your Pods, as Google Cloud does not route PUPI addresses to the internet or to other VPC networks by default.
It enables you to use VPC Network Peering to connect your GKE cluster to other VPC networks that use different PUPI addresses, as long as you enable the export and import of custom routes for the peering connection.
It preserves the fully integrated network model of GKE, where Pods can communicate with nodes and other resources in the same VPC network without NAT.
The options that you need to select when creating a private GKE cluster with PUPI addresses are:
Option A is incorrect because it does not use PUPI addresses for Pods, but rather RFC 1918 addresses. This does not solve the problem of address exhaustion in your enterprise. Option B is incorrect because it reuses the secondary address range for Services across multiple private GKE clusters, which could cause IP conflicts and routing issues. Option C is incorrect because it does not specify the options that are needed to create a private GKE cluster with PUPI addresses.