Essentially you can start up a new master node with the same IP/DNS details and it will assume the role of the original as the other answer aludes to, however the truth is that the master itself has no HA or fault tolerance.
If you were to experience an event which took a master and an indexer offline you would be left in an inconsistant state, as per Starfleets situation here; http://splunk-base.splunk.com/answers/65397/splunk-v5-clustering-and-ha
Its still a much improved HA solution and its pretty good to be fair, but there really needs to be better tolerance for the master failing 🙂
... View more