NDB High Availability uses multiple API Server VMs to provide the user interface and REST API services. The platform can automatically tolerate the failure of up to two service VMs, but if every API Server VM is offline, the API service becomes completely unavailable and manual intervention is required. Recovery requires bringing at least one API Server VM back online. Once an API Server is running, the NDB API service can restart and the remaining HA components can rejoin or recover through the normal resiliency mechanisms. Restarting HAProxy first cannot restore an API service when no API Server is available. Repository VMs should not be recreated because they contain NDB metadata, and registered Nutanix clusters do not need to be registered again. Therefore, the correct answer is A. NDB HA failure and recovery scenarios

================