Nutanix security and network-segmentation documentation clearly states that the segmented DSIP and the regular cluster DSIP are different entities, and that Prism Central needs a cluster DSIP in the same subnet to upgrade. In other words, placing the DSIP only in a secondary non-routable segment can break Prism Central’s ability to perform upgrades if Prism Central cannot reach the cluster using the expected subnet path. The documented fix is to maintain or configure a DSIP in the original communication subnet that Prism Central can use. That makes C the correct answer.

The other options are not the Nutanix-documented solution. Proxy ARP is not the prescribed remedy here, and disabling segmentation is unnecessary if the communication requirement can be satisfied by adding the proper DSIP. Uplink redundancy has nothing to do with the management-plane reachability issue in the question. So the authentic fix is to configure an additional DSIP in the original subnet used for Prism Central communication, which is C.