executor: sync deletable columns to binlog when remove record

[tangenta]

What problem does this PR solve?

Issue Number: close #53133

Problem Summary:

Previously, when TiDB writes delete records to binlog, column count of schema and data mismatch. This is because only public columns are selected, while data includes all deletable columns.

What changed and how does it work?

Make schema use all deletable columns.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
  1. deploy two TiDB clusters, and start pump & drainer.
  2. modify upstream TiDB code:

  diff --git a/pkg/config/config.toml.example b/pkg/config/config.toml.example
  index c1438a9e85..b73828367d 100644
  --- a/pkg/config/config.toml.example
  +++ b/pkg/config/config.toml.example
  @@ -416,7 +416,7 @@ capacity-mb = 1000.0
   # enable to write binlog.
   # NOTE: If binlog is enabled with Kafka (e.g. arbiter cluster),
   # txn-total-size-limit should be less than 1073741824(1G) because this is the maximum size that can be handled by Kafka.
  -enable = false
  +enable = true
   
   # WriteTimeout specifies how long it will wait for writing binlog to pump.
   write-timeout = "15s"
  diff --git a/pkg/ddl/column.go b/pkg/ddl/column.go
  index 206b64d513..1727ba4c44 100644
  --- a/pkg/ddl/column.go
  +++ b/pkg/ddl/column.go
  @@ -173,6 +173,7 @@ func onAddColumn(d *ddlCtx, t *meta.Meta, job *model.Job) (ver int64, err error)
                  // reorganization -> public
                  // Adjust table column offset.
                  failpoint.InjectCall("onAddColumnStateWriteReorg")
  +               time.Sleep(20 * time.Second)
                  offset, err := LocateOffsetToMove(columnInfo.Offset, pos, tblInfo)
                  if err != nil {
                          return ver, errors.Trace(err)

  3. execute SQL at upstream:

  -- session 1
  CREATE TABLE `t` (
    `a` int(11) DEFAULT NULL,
    `b` int(11) DEFAULT NULL,
    `c` int(11) DEFAULT NULL,
  ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin
  insert into t values (1, 1, 1), (2, 2, 2), (3, 3, 3);
  alter table t add column d int default 3;
  

  -- session 2
  delete from t where a = 1;
  

  4. execute SQL at downstream:

  mysql> select * from t;
  +------+------+------+------+
  | a    | b    | c    | d    |
  +------+------+------+------+
  |    3 |    3 |    3 |    3 |
  |    2 |    2 |    2 |    3 |
  +------+------+------+------+
  2 rows in set (0.00 sec)
  

  Even if we synchronize a 4-columns row to binlog, drainer can handle it correctly by decoding necessary(public) columns:
  https://github.com/pingcap/tidb-binlog/blob/6fba4f883a0c6c7e0d68b3b8aef1a53d9454b273/drainer/translator/mysql.go#L88-L90
• No need to test

  • I checked and no code files have been changed.

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

None

[tiprow]

Hi @tangenta. Thanks for your PR.

PRs from untrusted users cannot be marked as trusted with /ok-to-test in this repo meaning untrusted PR authors can never trigger tests themselves. Collaborators can still trigger tests on the PR using /test all.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 74.6091%. Comparing base (bdb0545) to head (23ea05f).
Report is 1139 commits behind head on master.

Additional details and impacted files

@@               Coverage Diff                @@
##             master     #53617        +/-   ##
================================================
+ Coverage   72.4359%   74.6091%   +2.1732%     
================================================
  Files          1506       1506                
  Lines        430579     433369      +2790     
================================================
+ Hits         311894     323333     +11439     
+ Misses        99367      90086      -9281     
- Partials      19318      19950       +632     

Flag	Coverage Δ
integration	49.1612% <66.6666%> (?)
unit	71.4387% <100.0000%> (+0.0265%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
dumpling	53.9957% <ø> (ø)
parser	∅ <ø> (∅)
br	50.4674% <ø> (+9.0192%)	⬆️

[wjhuang2016]

/retest

[D3Hunter]

suppose we are dropping/lossy-modify a column, this list won't contains the changed column while binlog think the column exist, not sure if binlog can handle it.

[tangenta]

For lossy modifying a column, it consists of two steps:

1. add a hidden column
2. drop the old column

Since DDL is only synced when the job state is done, the hidden column does not exists in downstream.

[tangenta]

Consistent with

tidb/pkg/planner/core/logical_plan_builder.go

Lines 4893 to 4895 in afd2343

	} else if b.inDeleteStmt {
	// DeletableCols returns all columns of the table in deletable states.
	columns = tbl.DeletableCols()

[D3Hunter]

when add column this list will contain the new column that binlog does not know now, can binlog handle it?

[tangenta]

Drainer can handle it correctly by only decoding necessary(public) columns:
https://github.com/pingcap/tidb-binlog/blob/6fba4f883a0c6c7e0d68b3b8aef1a53d9454b273/drainer/translator/mysql.go#L88-L90

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: D3Hunter, wjhuang2016

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [D3Hunter,wjhuang2016]
• pkg/ddl/OWNERS [wjhuang2016]
• pkg/table/OWNERS [wjhuang2016]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2024-05-28 13:27:07.85031357 +0000 UTC m=+2782981.607449143: ☑️ agreed by wjhuang2016.
• 2024-05-29 07:01:13.948109721 +0000 UTC m=+2846227.705245293: ☑️ agreed by D3Hunter.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.5: #53647.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-6.5: #53700.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-8.1: #54942.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.1: #56887.