The suggested answer is B, D.
If the Secondary VM is on an overcommitted ESXi host, using vMotion to migrate the Secondary VM to a different ESXi host can help alleviate the resource contention and improve performance. Additionally, turning off and turning on Fault Tolerance will recreate the Secondary VM, potentially on a different and less overcommitted datastore. These actions directly address the problem described.