Process Failover in a Cluster

[Gregg Fiehler]

I have an erlang cluster and a distributed application, this application creates processes for processing data.  Normally I have sibling processes created on each node since the data processing is unordered, so if a process or node goes down, the other two continue to process the data.  However, I have a new data source that must be processed in order so it will require a single process to manage this (yes there are other techniques, but in this case we want to single thread the processing).  Is there anyway to have the process restarted on another node if the node fails automatically?  Can I listen for process failure and node failure events on the other nodes and somehow determine which node should take over?  I basically need something similar to rabbitmq master slave queue failover, however, I won't have an active process on the other nodes.

Thanks

Gregg