The solution of the key-value conflict problem of MySQL bidirectional synchronization sharing _mysql

Problems that occur (multiple-master self-growth ID duplicates)
1: First we pass the test table structure of A,B
2: Off A, in B on the data table test (there is a self-growth ID) to perform an insert operation, return the insertion ID of 1
3: After stopping B, in a on the data table test (there is a self-growth ID) to perform the insert operation, the return of the insertion ID is also 1
4: Then we start the a,b at the same time, the primary key ID duplicate will appear

Workaround:
All we have to do is make sure that the growth data inserted on both servers is different.
such as: A Chachi number id,b interpolation even ID, of course, if more than the server, you can define the algorithm, as long as the difference can be

Here we add the parameters on a, b to implement the parity insertion

Add Parameters on A:MY.CNF

Auto_increment_offset = 1
Auto_increment_increment = 2
Such a auto_increment field produces a value of: 1, 3, 5, 7, ... Odd ID.

Add Parameters on B:MY.CNF

Auto_increment_offset = 2
Auto_increment_increment = 2
In this case, the auto_increment field of B produces a value of: 2, 4, 6, 8, ... Equal even ID.

As you can see, your auto_increment field is never duplicated between different servers, so there's no problem with the master-master structure. Of course, you can also use 3, 4, or N servers, as long as the guarantee auto_increment_increment = N again set the Auto_increment_offset for the appropriate initial value on it, that, Our MySQL can have dozens of primary servers at the same time without duplication of growth IDs.