I'm wanting to do a rather convoluted replication setup temporarily to
try and avoid a major switch over that might result in a certain
amount of down time.
Our current system is set up with master-master replication. I've
installed two new servers with a master-slave replication setup.
I've read that chained replication is simply done by enabling
log_slave_updates, but what I'd like to do is this:
old master <-> old master -> new master -> new slave
Leave the old masters in circular replication, but have one of them
replicate to the new master which then replicates to the new slave.
If I enable log_slave_updates on the old master that will replicate to
the new master will this result in duplicates back to the other old
master or will server-id matching prevent this happening? I recall
reading something about the server-id being used to track whether an
update should be done or not.
This is just a temporary measure to keep the two new servers in sync
with the old ones while we switch our websites and mail servers over
to the new MySQL servers. Once that is done I'll be removing the two
old MySQL servers from service.
Don't just do something...sit there!