I'd like to migrate our management server of our cluster (2 data, 2 SQL, 1 mgm) to a different physical host. Is it possible to do this with one rolling restart (RR) of the cluster?
Based on the documentation it seems that I have to done this using at least two RR. First add another ndb_mgmd on the new host, preform the rolling restart (with new configuration), stop the old host and perform another rolling restart (again with new configuration).
If so, do I have to use --initial when adding new management node or it will automatically get the configuration from the old management node cache?
Or is it possible to change physical host of the management node during one rolling restart? Because as soon as I change the host of the management the whole cluster will be without ndb_mgmd. Can't this cause problems during rolling restart?
Version (mysql-5.1.47 ndb-7.1.5), Debian 6.0.1.
