Currently, REI3 only supports integer primary keys. And yes, if you are running the DB in a cluster with multiple nodes, the IDs would be the same - but this is how it should be. Each cluster node needs to know the correct (eg. the same) ID for a specific record, otherwise your cluster would make no sense.
UUIDs are really useful to synchronize data between different DB systems (not nodes within a cluster, but separate systems). For this purpose, REI3 includes UUID attributes, which can auto-generate and be used to uniquely identify records between different systems.
A cluster needs to replicate / synchronize the same data - if your keys look different, you have a problem.
