Enable REPLACE INTO engine and Fix Foreign key locking issue

[harshit-gangal]

Description

While adding REPLACE INTO support for foreign keys in unsharded tables, we discovered that select queries executed for foreign key cascade were not able to acquire an adequate level of lock on the unique key due to missing gap locks. This led to incorrect results, and further plan execution would result in missing cascading rows, leading to incomplete data in the binary logs.

VReplication workflow relies on these binary logs for tasks like online-ddl or change log streaming.

Example case:

drop table if exists some_table;
create table some_table (
    id bigint,
    col varchar(10),
    primary key (id),
    unique index(col)
) ;

insert into some_table(id, col) values (3, null), (4, 5);

-- Session 1
begin;
select col from some_table where col in (5) or id in (3)  for update;


-- Session 2
begin;
select col from some_table where col in (('5')) for update;
-- This should block


-- Session 1
delete from some_table where col in (5) or id in (3) ;
insert into some_table(id, col) values (3, 5);
commit;


-- Session 2
-- Unblocked but returns 0 rows
select col from some_table where col in (('5')) for update;
-- This repeatable query in Session 2 now returns 1 row
revert;

To address the issue of blocked queries not providing updated results, we tried several techniques:

1. Locking parent rows with verification logic.
2. Using multiple select statements to acquire the necessary lock.
3. Employing a union statement for the lock.
4. Implementing a self-join query to obtain the lock.

These methods worked with two threads but led to incorrect/missing binary logs in a multi-threaded system.

As a solution, we're using the NOWAIT lock to promptly acquire the lock for cascade selection. NOWAIT ensures immediate lock acquisition or failure, which may result in more foreign key-related DMLs failing, necessitating query or transaction rollback. This approach, however, effectively addresses the problem of lock waiting and prevents incorrect results.

Related Issue(s)

• RFC: Foreign Key Support in Vitess #12967

Checklist

• Tests were added or are not required
• Did the new or modified tests pass consistently locally and on the CI

[vitess-bot]

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

• Ensure that the Pull Request has a descriptive title.
• Ensure there is a link to an issue (except for internal cleanup and flaky test fixes), new features should have an RFC that documents use cases and test cases.

Tests

• Bug fixes should have at least one unit or end-to-end test, enhancement and new features should have a sufficient number of tests.

Documentation

• Apply the release notes (needs details) label if users need to know about this change.
• New features should be documented.
• There should be some code comments as to why things are implemented the way they are.
• There should be a comment at the top of each new or modified test to explain what the test does.

New flags

• Is this flag really necessary?
• Flag names must be clear and intuitive, use dashes (-), and have a clear help text.

If a workflow is added or modified:

• Each item in Jobs should be named in order to mark it as required.
• If the workflow needs to be marked as required, the maintainer team must be notified.

Backward compatibility

• Protobuf changes should be wire-compatible.
• Changes to _vt tables and RPCs need to be backward compatible.
• RPC changes should be compatible with vitess-operator
• If a flag is removed, then it should also be removed from vitess-operator and arewefastyet, if used there.
• vtctl command output order should be stable and awk-able.

[GuptaManan100]

Rest LGTM!