Does 'run away' count as a recommendation? ;o)
I have master<->master replication set up (it's for a load-balanced helpdesk
system) and am currently trying to change back to master<->slave. When it
worked it was fine, but when it broke it caused chaos. The reason is that
when replication broke (due to a duplicate db entry) it took us two days to
realise what was going on. Both dbs were acting as 'masters' and processing
data individually so it wasn't immediately obvious there was a problem. We
ended up with helpdesk tickets that had the same ticket number but different
contents on each server. So when we decided to restore we had to pick one
db and lose the data from the other. We're still having lots of problems.
Unless you have a way of monitoring when replication goes down then
master<->master has the potential a create a real mess. I'm very new to
mysql but that's been my experience.
Christian Schramm-5 wrote:
> Hi @all,
> I've created a simple setup with 2 servers, each acting as master and as
> So it does not matter on which server you will write.
> Of course you may get problems with duplicate entries and depending on
> the server performance also a synchronisation lag.
> Does anyone in the list have experience with multi master setups like
> Any recommendations?
> I want to realize the current project i'm working on with replication
> because the database is not optimized to run e.g. on a ndbcluster.
> Mit freundlichen Grüßen / Kind regards
> Christian Schramm
View this message in context:
Sent from the MySQL - Replication mailing list archive at Nabble.com.