[MySQL-replication] Master to master replication

Database System wrote:
Hi,

On a master to master replication, if a table doesn't have a auto_increment column, how the server(s) prevents update statement from executing twice on same server.

For example, one update is done on server A, and another update is done on server B. But these two update statements are on both binary logs of server A and server B. When server A reads binary log from server B to replicate, two update statement are in binary log, how server A only execute one update done on server B and ignore another one that was done on server A?

Thanks,
Lisa

-----Original Message-----
From: Saravanan
Sent: Tuesday, March 31, 2009 2:43 PM
To: database100@yahoo.com; Mats Kindahl
Cc: replication@lists.mysql.com
Subject: Re: Master to master replication

Hi,
There are two types of logging in replication. Binary log and
Relay log. Binary log is the log of update statements the
server updated. The second one is log that is updates
received from other server.
For example server A will not read the binary logs it updates
and it will write in the relay logs that it reads from server B.
In Master to Master setup don't enable log_slave_updates.
Thanks,Saravanan

--- On Wed, 4/1/09, Mats Kindahl wrote:

From: Mats Kindahl <mats@sun.com>
Subject: Re: Master to master replication
To: database100@yahoo.com
Cc: replication@lists.mysql.com
Date: Wednesday, April 1, 2009, 3:59 AM

Database System wrote:
a auto_increment column, how the server(s) prevents update
statement from executing twice on same server.
update is done on server B. But these two update statements
are on both binary logs of server A and server B. When server
A reads binary log from server B to replicate, two update
statement are in binary log, how server A only execute one
update done on server B and ignore another one that was done
on server A?
Hi Lisa,

Each server have to have a server id assigned, so statements
executed on A will
be tagged with the server id of server A (and the same for
B). When A then reads
an event (statement) from itself, it will just skip that
event and move on to
the next one.

Best wishes,
Mats Kindahl


--
Mats Kindahl
Senior Software Engineer
Database Technology Group, Replication team
Sun Microsystems

--
MySQL Replication Mailing List
For list archives: http://lists.mysql.com/replication
To unsubscribe:
http://lists.mysql.com/replication?unsub=suzuki_babu@yahoo.com

Rick James wrote:
But, if you have Slaves hanging off both Masters, you need log_slave_updates, right?

Rick James
MySQL Geeks - Consulting & Review

From: Rick James <rjames@yahoo-inc.com>
Subject: RE: Master to master replication
To: "Saravanan" <suzuki_babu@yahoo.com>, database100@yahoo.com, "Mats Kindahl" <mats@sun.com>
Cc: replication@lists.mysql.com
Date: Friday, April 3, 2009, 1:29 PM
But, if you have Slaves hanging off both Masters, you need
log_slave_updates, right?

Mats' answer may be less efficient, but still holds.


Rick James
MySQL Geeks - Consulting & Review


doesn't have
another
http://lists.mysql.com/replication?unsub=suzuki_babu@yahoo.com
--
MySQL Replication Mailing List
For list archives: http://lists.mysql.com/replication
To unsubscribe:
http://lists.mysql.com/replication?unsub=database100@yahoo.com

Database System wrote:
My understanding is, whether log_slave_updates enabled or not, it doesn't affect master to master replication. Anybody has different opinion?

Lisa


--- On Fri, 4/3/09, Rick James wrote:



--
MySQL Replication Mailing List
For list archives: http://lists.mysql.com/replication
To unsubscribe: http://lists.mysql.com/replication?unsub=mats@sun.com