Ed Pauley II wrote:
This is another geographical location with automatic failover if there is a problem, network, hardware etc. with the primary location. When the problem is corrected, or corrects itself the traffic is automatically sent back to the primary location. Without 2-way replication data would be lost. We have been doing this for since MySQL 4.0 was released.
Ok, If that's your problem, they might have something to you in the near future with the new master election. It'll probably work on a 2-way replication when the old master is restored.
While this is not out yet, what you might do is what I did back with 3.23 which is: bring old restored master up as slave, wait for all changes replicate, shutdown temp master, set it as slave, bring old master as master and bring temp master as slave, restoring production scheme.
It is not a multi-master setup. The master at each location is both master and slave to each other. The slaves are only slaves to the master in their respective locations. My problem is really with how to load balance the slaves at each location.
Load balance them using the local master or using both slaves as a load balance ?
I'd rather go for the first one because the second one will raise more problems than solve, specially related to network delays.
If, in the first case, one master fails you can issue a "change master" to the other master and it'll keep running. As you'll only have the slave there will be no load balancing on that site.
--renato -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
