On 30/06/2005 5:23 p.m., Simon Garner wrote:
We do something similar. While you can mix MyISAM, InnoDB and NDB tables
in one database, and you can use --replicate-wild-do-table to replicate
only the non-clustered tables, we found this to be non-ideal as you then
either have to name your tables with some prefix to indicate if they are
clustered or not, or manually update your replicate-wild-do-table
configuration every time you add or remove a table.
I forgot to say, what we decided to do for now is to store the tables in
two separate databases, e.g. 'main' and 'cluster'. Then you can use
--replicate-wild-do-table=main.* to replicate the MyISAM/InnoDB tables
in 'main' and keep your NDB tables in the 'cluster' database. You can
still join across databases, e.g. SELECT t1.*, t2.* FROM main.t1,
cluster.t2, so the only disadvantage of this is your application may
need modifying to refer to tables with their database name.
-Simon
6 messages in com.mysql.lists.clusterRe: questions about ndb
