Thank you for the reply. I thought that if you deleted the auto.cnf and restarted the server you automatically get appropriate UUID regardless whether it is inside or outside VM. Am I correct here?

Best regards,

Jacob Nikom

You only have to edit the auto.cnf file if you copied a VM with MySQL already installed. Since there will be an existing auto.cnf file on the new VM, you will need to edit that file so that both of the VM’s don’t have the same UUID. On a fresh install of MySQL, you are correct in that a new auto.cnf file will be generated.

Thanks,

Tony

Thank you for the great article. This is one of the clearest descriptions of the MySQL installation that I have ever read.
One small correction about auto.cnf file – you don’t have to edit it manually, it will be generated automatically if there is no auto.cnf file in the corresponding data directory. If auto.cnf file already exists, the server will leave it the same.

Best regards,

Jacob Nikom

If there is an existing database on the master.
Make sure that you have restored an exact copy of the master into the slave.

on master
mysqldump -u USERNAME -p DATABASE > dump.sql
copy dump.sql to the slave

on slave
mysqldump -u USERNAME -p DATABASE < dump.sql

now any record you add on the master will show on the slave.

I am not sure what you are asking for question number one.

I’m trying to get this to work between our development database and our production database. We make changes to development and then before pushing the code changes I need to verify that the database are the same and all schema changes have been updated on production

1. Is it possible to pass a test when the auto-increment changes. Many tables are failing due to the auto-increment values changing. This is fine for our needs since all I want to check for is new tables and changes to table schemas

2. Is it possible to run this on databases with different names, ie… database_dev:database_prod – Same database just different names

You are correct. If you are creating a slave from a database where the binary logs have been purged, then using the mysqlreplicate utility isn’t your best option. In this case, you would need to copy the data from the master to the slave, and then start GTID. You could copy the data with mysqldump or MySQL Enterprise Backup. This link explains how to use GTID’s if you already have a master/slave configuration.

Thanks.

It would be nice to see how to add a new GTID slave without –start-from-beginning. This does not seem to be mentioned anywhere (for example, the mysqlreplicate page has no mention of GTID).