During a FlashArray//XR5 upgrade, when should the Implementation Engineer run start_ndu controller-ndu to set the...

When executing a Hardware Non-Disruptive Upgrade (HWNDU) to a new chassis generation, such as moving to a FlashArray//XR5, the new controllers must be properly prepped to join the existing, older-generation cluster. The correct sequence of operations is critical to prevent the cluster from rejecting the mismatched hardware.

Once the new //XR5 controller is physically inserted into the slot of the removed older controller, the Implementation Engineer must first install the correct target version of the Purity//FA operating system locally using the pureinstall command via the KVM console. After Purity is successfully installed, but strictly before the controller is logically initialized to join the cluster (via the puresetup secondary command), the engineer must execute the start_ndu controller-ndu script.

This script applies specific internal HWNDU tunables—such as relaxing hardware mismatch checks, bypassing specific alert monitors, and adjusting heartbeat timeouts. Setting these tunables ensures that when the controller is finally initialized, the surviving primary controller safely accepts it as a valid secondary partner without triggering a system fault, hardware rejection, or disrupting active host I/O.