What is the expected outcome that the administrator must manage?

This question refers to the behavior of Nutanix Metro Availability in conjunction with a Witness VM during a site-to-site communication failure. Metro Availability provides a zero-RPO, synchronous replication solution. To prevent " split-brain " (where both sites try to serve the same data independently), the Witness VM acts as a tie-breaker. When the two clusters lose connectivity with each other, both will immediately attempt to contact the Witness to " acquire the lock " on the protection domain. The Witness is programmed to prioritize the original " Active " site. If the primary cluster can reach the Witness, it will successfully acquire the lock and continue serving I/O locally to the guest VMs, ensuring zero downtime for the applications. The secondary site, unable to reach the primary or acquire the lock from the Witness, will transition its storage container to an " Inactive " state to protect data integrity. The administrator must manage this state by ensuring the Witness is always reachable from both clusters via independent network paths. If the primary site cannot reach the Witness, it will stop serving I/O to prevent inconsistent data states, highlighting the critical importance of Witness availability in a high-availability BCDR design.