When performing cluster recovery, support specifying the database or filtering conditions

Thank you very much to the development team for their excellent work in providing s3-based point-in-time recovery (PITR) for backups and restores. However, one drawback is that it is currently not possible to specify the specific database or table for backup and restore operations.

The documentation I have referred to is available at:

In my cluster, multiple databases are being used simultaneously, and occasionally, due to inactivity, individual databases need to be restored to a previous point in time using PITR, while other databases remain unaffected.

It would be great if this functionality could be implemented, taking inspiration from the implementation of TiDbOperator, as described in:

I am eagerly looking forward to this feature, and once again, I want to express my gratitude to the development team for their efforts!

Hello @wang_jie,
Thanks for the feature request. Just to let you know, single table as well as single database restore is complex in MySQL. Even when done manually, you must extract each table from the backup, re-create the schema in the running MySQL server, drop the tablespace, copy the backup tablespace, import that tablespace, then replay binary logs. Unfortunately, replaying binary logs does not support filtering at a per-table basis. PITR in that sense just isn’t possible. PITR for the entire MySQL server is possible and so that’s what our process does.