Hi,
George lin Wrote:
-------------------------------------------------------
> Dear MySQL Cluster pioneers:
>
> According to the official documents of :
> 1,
>
https://dev.mysql.com/doc/refman/5.7/en/mysql-clus
> ter-nodes-groups.html
> 2,
>
https://dev.mysql.com/doc/refman/5.7/en/mysql-clus
> ter-online-add-node.html
>
> I have some questions for online adding new data
> nodes for over hundreds PB data:
>
> I, how many nodegroups should be created in a 1
> replica cluster when adding 2 data nodes or 4 data
> nodes ? since replica is 1, so one should create 2
> nodegoups for 2 data nodes and 4 nodegroups for 4
> data nodes, am I right?
>
Right
> II, if just adding new data nodes and without
> changing other datanode settings, such as not
> changing "1 replica", according to
>
https://dev.mysql.com/doc/refman/5.7/en/mysql-clus
> ter-online-add-node-basics.html , could one skip
> the 7th and 8th steps to not reorganize the old
> existing tables and data, to let the old tables
> and its old-and-new data to stay on the old
> computers, but let the latter creating tables and
> its new data only to spread to new computers?
> In a word, is it work stably to only follow the
> before 6 steps to add a new nodegroup with
> regorganize the old tables?
If you don't reorganize the old tables they will
remain on the old computers. New tables will spread
on all data nodes. So this means data will be
unevenly spread without reorg.
There are some parameters to decide how to spread
the data in a table. But this isn't the standard.