workaround xtrabackup error with old_alter_table

[deepthi]

Description

Some of the xtrabackup tests have been flaky for a while. Today @vmg was able to extract the actual error from a failing test.
This seems to be a known issue with a documented workaround. Implementing the workaround.
https://www.percona.com/blog/2017/08/08/avoiding-the-an-optimized-without-redo-logging-ddloperation-has-been-performed-error-with-percona-xtrabackup/

The blog post has 3 suggestions:

• --lock-ddl : only works with Percona MySQL server, we are running community MySQL server in CI
• --lock-ddl-per-table: did not fix the problem
• old_alter_table: 🤞

Related Issue(s)

Fixes #8315

Checklist

• Tests were added or are not required
• Documentation was added or is not required

Deployment Notes

[aquarapid]

Do we also want to put a note somewhere in https://vitess.io/docs/user-guides/operating-vitess/backup-and-restore/creating-a-backup/ ?

[systay]

Lets make sure this should not go into an earlier branch first

[shlomi-noach]

Noting that enabling --old_alter_table effectively disables concurrent DML on table while ALTER TABLE takes place. We do happen to offer Online DDL in vitess, but users who will run direct ALTER TABLE and with ALGORITHM=INPLACE LOCK=NONE may be surprised by this change of behavior.

[vmg]

@shlomi-noach afaict this change of behavior is only on the end-to-end tests and doesn't apply to normal production usage.

[shlomi-noach]

this change of behavior is only on the end-to-end tests

Ah, then please just ignore me.

[shlomi-noach]

Gonna resubmit these changes into release-9.0

[shlomi-noach]

Resubmitted against release-9.0: #8340

[systay]

Closing so we don't accidentally hit that big green button