Optimize MYSQL_REPLICATION_SLAVE_DUMP behavior

[mostafa-rz]

Name and Version

bitnami/mysql:8.4-debian-12

What architecture are you using?

amd64

What steps will reproduce the bug?

1. Enable the MYSQL_REPLICATION_SLAVE_DUMP
2. Start replica
3. During the dump process, insert new values
4. The MySQL table LOCK is not working and it is allowed the insert
5. The database snapshot is not consistent anymore

What is the expected behavior?

Writing to the database must not be allowed during the dump process.

What do you see instead?

Based on the MySQL documents:

Warning
Leave the client from which you issued the FLUSH TABLES statement running so that the read lock remains in effect. If you exit the client, the lock is released.

The problem relies on this part of the code. The FLUSH TABLES WITH READ LOCK; is executed by a client and afterward it exits, therefore the lock is released.

I will make a pull request that mysqldump take responsibility for locking the tables and finding the binlog position.

Additional information

No response

[carrodher]

Thank you for opening this issue and submitting the associated Pull Request. Our team will review and provide feedback. Once the PR is merged, the issue will automatically close.

Your contribution is greatly appreciated!