Targeting a single database with these options:
mysqldump -u MY_DB_BACKUP_USER -p MY_DB_NAME --add-drop-table --add-drop-database > my_db_name_mmddyyyy.sql
will successfully backup the database, but if you were to restore the database backup file back into a database with a newer schema (additional tables for example) like so:
mysql -u my_db_access_user -p MY_DB_NAME < my_db_name_mmddyyyy.sql
you would end up with a “merged” version of the database that would have new tables from the more recent schema and restored tables from the backup:
I learned this the hard way recently when attempting to restore a database to an earlier version following a failed Redmine 3.0.2 upgrade. While this didn’t break anything (other than maybe a future upgrade attempt that doesn’t properly handle existing tables), it wasn’t what I intended.
Going forward, if I want to backup a single database and have the backup file contain the statement necessary to first drop the database before restoring it (a full restore for example), I’ll need to call the
mysqlbackup command like so:
mysqldump -u MY_DB_BACKUP_USER -p --databases MY_DB_NAME --add-drop-table --add-drop-database > my_db_name_mmddyyyy.sql
--databases statement. If you didn’t know this, I hope it helps you from making the same mistake.