You are deploying new workloads on a GKE Autopilot mode cluster. You need to ensure that the Pods are scheduled on nodes that use Arm architecture CPUs. The cluster currently has no Arm- based CPUs. Y

Question

Options

• AApply Pod tolerations to request GKE to avoid scheduling Pods on nodes that do not have Arm-
• BApply node taints on node pools to tell GKE to only schedule your workloads on Arm-based CPU
• CDeploy a new cluster in GKE Standard mode, and set up a node pool with Arm-based CPU
• DRequest the Scale-out compute class and the arm64 architecture in your manifest.

• A
  6% (2)
• B
  3% (1)
• C
  19% (6)
• D
  72% (23)

Explanation

GKE Autopilot abstracts away node pool management-you cannot manually create or configure node pools. To request specific hardware in Autopilot, you declare compute requirements directly in the Pod manifest using compute classes and node selectors. Requesting the scale-out compute class with the arm64 architecture tells GKE Autopilot to automatically provision the appropriate Arm-based nodes to satisfy the Pod's requirements. This requires zero manual cluster operations. Options A and B (tolerations and taints) are node-pool-level concepts that only apply in GKE Standard mode. Option C requires switching to Standard mode, which increases operational overhead and contradicts the goal of minimizing cluster operations.

No community discussion yet for this question.