I am trying to get a little clarification on how the master/replica process works. My configuration is a 4-server cluster, with one master and 3 read-only replicas.
Which of these is correct for a Java server-side update script on the master that is batch updating many records at once?
During the session, after each record is updated on the master copy of the database, it immediately updates the replica’s data.
At the beginning of the session, there is a lock on the copy of database the master has, the update occurs, then once the session is closed, the replicas are updated.