A: vSphere 5.0's complete rewrite of vSphere HA clustering introduces new ways in which its Fault Domain Manager (FDM) agents on each ESXi host communicate with each other. Prior to this version, vSphere HA used a primary/secondary approach to manage agent roles on each host. This architecture changes in vSphere 5.0 to a master/slave approach. One agent in each vSphere HA cluster is chosen to assume the role of master and all others assume the role of slaves.
The master is determined through an election process. That process occurs any time vSphere HA is enabled, and when an existing master host fails, is shut down, or is placed into management mode. It can also occur when a management network partition occurs.