Solving problems with mysql replica (no replies)

HI everybody,

I have had some problems with MySQL's replica. I have two servers in master-master architecture, so, I have configured a master and a slave in both. When I have problems in the replica , in some cases, I have to do a lock of tables in master A, make a backup, take note about mysql-bin and position and then I have to restore backup on my master B. Last step is unlock tables.

I think that this is the normal way to restore replication when an error occurs, but if in both masters there are read/write operations, I have got a problem!

Are there any way to restore replication between masters but without stopping or blocking write operations? I know 'SET GLOBAL sql_slave_skip_counter=1' restarting slave, but I think that this is not good.

Any ideas about how can I solve problems in MySQL's replica? I need a good protocol to solve it.

Maybe, if both masters there are active and receive read/write operations, shoud I to configure another architecture? Obviously different, like as cluster or whatever.

I hope your advices.

Best regards,

Toni.