MySQL Cluster

[John Nemeth <jnemeth%cue.bc.ca@localhost>]

     I've been working on packaging MySQL Cluster.  This is the
distributed version that has redundancy (master-master replication)
and failover.  At this point, I have one large package.  It compiles
and early testing looks good (I'm still reading documentation and
playing with it).  Most of the MySQL packages currently have a
client and a server part.  I'm assuming that people would like to
see that seperation maintained?

     On the server side, MySQL Cluster has what it calls "nodes".
These are seperate pieces that are intended to be run on seperate
systems (although they can be run on the same).  The nodes perform
different functions:  a management node that co-ordinates everything
and provides configuration information to the other nodes, an SQL
node (mysqld) which performs the SQL server function similar to
traditional MySQL, and a data node where the actual data is stored.
Is there a strong preference to see the server components split
into seperate packages, or would it be okay to do all the server
pieces as one package?  Note that I haven't investigated how hard
it would be to do this yet.  If it complicates things too much, I
might do it as one package anyways.