Web lists-archives.com

Re: Pointers to Mysql Replication




On 7/5/2015 10:48 PM, Claudio Nanni wrote:
Hello Jatin,

> We are basically intending to implement High Availability for our application. In this regard , we want to use Mysql replication for this purpose.

Your scenario(MySQL Master/Slave with Failover/Failback procedures) has been a typical one for ages now, so I am sure many good advices about it will arrive.

While, I just want to bring into the table the possibility to use Galera replication which is synchronous*, multi master, and does not need failover/failback procedures, nodes syncs and desyncs automatically. Indeed I believe that this simple Master/Slave traditional scenario can represent an interesting (base case) application for Galera, I'd like other expert opinions on this too.

Galera is of course a bit more complex so this has also to be taken into account.

*(It has an optimistic multinode transactional approach so that in some cases a transaction might be rolled back in case of conflicts.)

Cheers
--
Claudio
Thanks Claudio,

I think having a master-master replication is what i precisely need. Basically if my application writes on the DB on node-1 it should be replicated to the DB running on node-2. After a failover my application runs on node-2 and when it writes to the DB on node-2 it should be replicated to the DB on node-1.

I have a question here though,

Say if i am using a DB instance for quiet some time on node-1 and it has data in it. If i decide to have replication with another DB instance , will all the data from node-1 be first replicated to node-2's DB ? Or will it just start the DB replication from the point where i configured DB replication ?

Thanks
Jatin