hi colin! galera cluter have something like it?

2013/9/23 Colin Charles byte@mariadb.com

On 24 Sep 2013, at 06:32, Roberto Spadim roberto@spadim.com.br wrote:

...

Hi guys, i got a question yesterday... there's an async cluster solution? example... multi-master and async? if it's async, it can't be multi-master, i'm

wrong?

...

You do say "async cluster solution", so I'm wondering if you mean NDBCLUSTER? If you are, you might refer to:

http://www.clusterdb.com/mysql-cluster/mysql-cluster-asynchronous-replicatio...

Naturally, the above doesn't apply to MariaDB as we don't ship NDBCLUSTER.

-- Colin Charles, Chief Evangelist MariaDB | t: +6-012-204-3201 | Skype: colincharles

Hi Colin, hi guys! =) other doubt, but not a "cluster" doubt, now a replication doubt...

check if this is possible, or i'm getting crazy =)

two mariadb servers (last version ~10.0.5): A and B A have the same tables of B B have the same tables of A they are both empty

when i write at A it replicate to B when i write at B it replicate to A

if i have a solution at client site (application) that allow write to A only using a primary key starting with "A" and writes at B a primary key starting with "B", in other words..

writing at A: using WHERE pk_field="A" writing at B: using WHERE pk_field="B"

could this replicate A to B, and B to A? for example, does this get in a "infinite loop", like:

A: UPDATE table SET f=1 WHERE pk_field="a" A replicate to ->B: UPDATE table SET f=1 WHERE pk_field="a" B replicate to ->A: UPDATE table SET f=1 WHERE pk_field="a" A replicate to ->B: UPDATE table SET f=1 WHERE pk_field="a" endless loop replicating from A->B, B->A, A->B, B->A, etc...

this could happen, or replication know when it do a "loop" at replication? i'm not sure if i'm right, but since A is master of B, and B act as a slave for first update, i think it will not replicate to A, i'm right?

thanks!

Hi Hartmut! i will read this docs, i have some replications and multi source replications, but i don't have a setup like this :) i will test it! thanks :D !

2013/9/24 Hartmut Holzgraefe hartmut@skysql.com:

On 09/24/2013 08:41 AM, Roberto Spadim wrote:

...

this could happen, or replication know when it do a "loop" at replication? i'm not sure if i'm right, but since A is master of B, and B act as a slave for first update, i think it will not replicate to A, i'm right?

that's why each participant in a replication setup needs to have a unique server_id so that it can identify replication events that originated from itself and ignore them.

Also a mysql server that acts as both replication master (writing its own binlogs) and slave (receiving binlog events from another master) will only forward replication events received from its master to its own binlog if log_slave_updates is enabled which is off by default.

See also:

https://dev.mysql.com/doc/refman/5.5/en/replication-options-binary-log.html#...

http://dev.mysql.com/doc/refman/5.5/en/replication-options.html#option_mysql...

-- Hartmut Holzgraefe, Principal Support Engineer (EMEA) SkySQL | http://www.skysql.com/

_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp