Update mariadb-galera.yaml

[aedan]

This forces the masters to be in sync for updated, deletes. This helps to prevent deadlocks as well

[BjoernT]

I don't think it will help with dead locks, in contrary it might introduce them especially on heavy inserts/updates.

During the check, the node blocks new queries while the database server catches up with all updates made in the cluster to the point where the check was begun.

Let's track the metrics such as Innodb_deadlocks and Table_locks_waited and Innodb_row_lock_waits to see how those parameters change the outcome of locking.

[aedan]

Well in testing. It definitely did help. I think the problem is that the errors are not actually dead locks. They are being reported as suck from openstack, but the problem is that a commit to the database is happening and then for whatever reason the connection is dropping. Then when it is reconnected. It tries the transaction again and it is already there. So the DB sync fails and will not complete. Is this the end all fix? No, but it does appear to help it out.

[aedan]

Ok. This final commit contains the changes that cloudnull and I discussed. I have tested it and have not seen any deadlocks or DB issues.