Point in Time options with logical/mysqldump? - Source

Revision: 109208
User: Ross M
Date: 2021-08-16 10:34

Answer

With mysqldump, incremental backups are the binary logs from the point in time when the consistent snapshot was taken.

To keep a record of this point in time look at the --master-data and --dump-slave options. Which to use depends if the binary logs are from the current replica, or from the master of the current replica. If using a single machine use --master-data.

When using these options always use --single-transaction to ensure its a consistent transaction.

To do this in parallel, a consistent snapshot of independently running mysqldumps is needed (per database, or individual table - with --single-transaction). To do this use [[flush-tables-for-export|flush table with read lock]] before starting all of the mysqldump instances.  Once started this all tables read lock can be released. Also see [[backup-lock]] to ensure that tables aren't changed during the dump resulting in an unusable backup. Only one of these needs to include --master-data/dump-slave. Recommend the =2 option to ensure that the information is in a comment. Because of MDEV-7611 (sigh, still not fixed) the recovery should have an invalid host/user/password during restore, and the correct this when starting to replicate again.

Parallel loading is loading these each in a separate session. Because the data is independent of the other dumps there's no conflict here.

When shipping binlogs use --start-position set to --master-data/dump-slave position, as an absolute position on where to start on the first file.

Content reproduced on this site is the property of its respective owners, and this content is not reviewed in advance by MariaDB. The views, information and opinions expressed by this content do not necessarily represent those of MariaDB or any other party.

Point in Time options with logical/mysqldump? - Source

Answer

Products

Services

Pricing

Resources

About Us

Download MariaDB