Nutanix Updated NCP-BC-7.5 Exam Questions and Answers by daisy-may

Replication in a Nutanix environment always begins with an " initial seed " . This first replication task is a full, 1-to-1 copy of all the data blocks currently occupied by the virtual machines on the source cluster. Unlike subsequent incremental snapshots—which only transfer the changed blocks—the initial seed requires sufficient physical or logical space at the destination to hold the entire data set of the protected VMs.

In this scenario, the administrator attempted to protect 4TB of virtual machine data but restricted the destination container ' s capacity to a mere 200GB. Because the destination container reports only 200GB of available space, the replication job will immediately stall or fail to progress as soon as the incoming data stream exceeds that limit. Nutanix replication requires a " landing zone " that can accommodate the full size of the initial snapshot before it can begin the process of data reduction (like compression or deduplication) at the remote site. To fix this, the administrator must increase the advertised capacity of the VMData container on Cluster B to at least 4TB (plus overhead for future growth and snapshots) to allow the initial full seed to complete successfully, thereby establishing the baseline for the BCDR protection schedule.

Establishing disaster recovery between an on-premises data center and a Nutanix Cloud Availability Zone (AZ) requires a robust communication path that can traverse different network boundaries. Unlike local replication within a single site, cross-site replication—especially to a public cloud environment—relies on the ability of the clusters to identify and reach one another over an external network. Nutanix Disaster Recovery requires that both the on-premises cluster and the Nutanix Cloud AZ instance have external IP addresses configured for their respective Controller VMs (CVMs) and virtual interfaces.

These external IP addresses allow the Cerebro service at the source site to establish a secure handshake with the Cerebro service at the cloud site. Without these routable external IPs, the replication traffic is unable to find its destination, leading to the immediate failure of the configuration as observed in this scenario. While identical hypervisors (Option A) are often preferred for simplicity, Nutanix supports Cross-Hypervisor DR (CHDR). Furthermore, deduplication status (Option C) does not prevent the establishment of a replication link. Synchronous replication (Option D) is restricted by latency requirements and is not a fundamental prerequisite for basic connectivity to a cloud AZ. Therefore, verifying the presence of external, reachable IP addresses is the mandatory first step in troubleshooting cross-site connectivity issues between on-premises and cloud environments.

Failback is a critical phase of the disaster recovery lifecycle where workloads are returned to their original primary site. When an " Unplanned Failover " is executed, the system assumes the primary site is gone or unmanaged. Once the primary site (AZ1) is restored, the administrator cannot simply " re-run " the failover in reverse without preparing the source site. The " Clean up of Entities " (Option C) is a mandatory step that must be performed on the original primary cluster after it returns to a healthy state. This process clears the stale VM registrations and metadata that existed prior to the outage, signaling to the Nutanix orchestrator that AZ1 is now ready to re-receive the workloads. Until this cleanup is completed, the Recovery Plan will not list the VMs as eligible for failback because it still detects the " ghost " of the VM at the destination. While reverse replication (Option B) is also necessary to prevent data loss during failback, the reason the VMs are entirely missing from the Recovery Plan interface during the initial failback attempt is the lack of entity cleanup. Performing this cleanup ensures a smooth transition back to primary production without metadata conflicts or registration errors.

When a replication job is marked as " ongoing " but the progress (bytes completed) is stagnant, it typically indicates a mismatch in the configuration or a failure in the orchestration logic rather than a total network blackout, especially when virtual IPs are confirmed reachable. In a " new setup, " a very common root cause is an incorrectly configured Protection Policy.

This can include issues such as mismatched category assignments, incorrect target site selection, or an invalid replication schedule that the system is unable to initialize properly. If the policy itself is misconfigured—for instance, if it is attempting to replicate to a storage container that hasn ' t been mapped or lacks permissions—the Cerebro service may initiate the task but then fail to progress the data transfer. While blocked ports (Options C and D) are common culprits in replication failures, they usually result in immediate " unreachable " alerts or job failures rather than a " stuck " status where connectivity is otherwise reported as healthy. If the Acropolis service were down at the remote site (Option A), the cluster would likely report the site as disconnected or show a critical service alert in Prism. Therefore, the most logical first area to investigate in a new, non-progressing setup is the logic and settings defined within the Protection Policy itself.