Storage: Fix DataTypePtr is not shared as expected by JaySon-Huang · Pull Request #9939 · pingcap/tiflash

JaySon-Huang · 2025-03-04T17:08:42Z

What problem does this PR solve?

Issue Number: close #9947

Problem Summary:

There are about 13,000 tables with 60 enum column in each table, turning into 780,000 DataTypeEnum instances. They takes about 4GB memory.

What is changed and how it works?

before_share_enum.svg.zip
after_share_enum.svg.zip

Storage: Fix DataTypePtr is not shared as expected
* Introduce a class `DataTypePtrCache` and manage the shared cache of `DataTypePtr` instances.
* Introduce `DataTypeFactory::getOrSet(const ASTPtr & ast)` and try to find the cache with data type name as "ast->range.first, ast->range.second"
logging: Turn the logging level of "updateTableColumnInfo" into debug because that could cause lots of logging when restarting tiflash

Check List

Tests

Unit test
Integration test
Manual test (add detailed scripts or steps below)

Create 36,000 table with schema like

CREATE TABLE IF NOT EXISTS enum_tbl_000 ( field1 VARCHAR(255), field2 VARCHAR(255), field3 VARCHAR(255),  enum1 ENUM('T', 'F'), enum2 ENUM('T', 'F'), enum3 ENUM('T', 'F'), enum4 ENUM('T', 'F'), enum5 ENUM('T', 'F'), enum6 ENUM('T', 'F'), enum7 ENUM('T', 'F'), enum8 ENUM('T', 'F'), enum9 ENUM('T', 'F'), enum10 ENUM('T', 'F'), enum11 ENUM('T', 'F'), enum12 ENUM('T', 'F'), enum13 ENUM('T', 'F'), enum14 ENUM('T', 'F'), enum15 ENUM('T', 'F'), enum16 ENUM('T', 'F'), enum17 ENUM('T', 'F'), enum18 ENUM('T', 'F'), enum19 ENUM('T', 'F'),  enum20 ENUM('pending', 'running', 'finished', 'failed'), enum21 ENUM('DEFAULT', 'ALL', 'PREDICATE', 'LIST'), enum22 ENUM('NONE', 'READ', 'INTEND', 'WRITE') ) DEFAULT CHARSET=utf8mb4;

Check the memory consumption without/with this fix. From 13:40 to 13:45, without this fix, tiflash consume 15GB memory. From 13:45 to 13:55, with this fix, tiflash consume 8GB memory.

No code

Side effects

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

Documentation

Release note

Fix the issue that TiFlash may consume lots of memory when there are many `ENUM` columns on TiDB

Signed-off-by: JaySon-Huang <tshent@qq.com>

dbms/src/DataTypes/DataTypeFactory.h

CalvinNeo

lgtm

Signed-off-by: JaySon-Huang <tshent@qq.com>

ti-chi-bot · 2025-03-06T07:59:24Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: JinheLin, Lloyd-Pottiger

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

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~OWNERS~~ [JinheLin,Lloyd-Pottiger]

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 · 2025-03-06T07:59:26Z

[LGTM Timeline notifier]

Timeline:

2025-03-06 02:37:58.64147587 +0000 UTC m=+495591.770395612: ☑️ agreed by Lloyd-Pottiger.
2025-03-06 07:59:26.07805223 +0000 UTC m=+514879.206971988: ☑️ agreed by JinheLin.

ti-chi-bot · 2025-03-06T08:49:01Z

In response to a cherrypick label: new pull request created to branch release-8.5: #9952.

close #9947 Storage: Fix DataTypePtr is not shared as expected * Introduce a class `DataTypePtrCache` and manage the shared cache of `DataTypePtr` instances. * Introduce `DataTypeFactory::getOrSet(const ASTPtr & ast)` and try to find the cache with data type name as "ast->range.first, ast->range.second" logging: Turn the logging level of "updateTableColumnInfo" into debug because that could cause lots of logging when restarting tiflash Signed-off-by: JaySon-Huang <tshent@qq.com> Co-authored-by: JaySon-Huang <tshent@qq.com> Co-authored-by: JaySon <tshent@qq.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com>

…ngcap#9952) close pingcap#9947 Storage: Fix DataTypePtr is not shared as expected * Introduce a class `DataTypePtrCache` and manage the shared cache of `DataTypePtr` instances. * Introduce `DataTypeFactory::getOrSet(const ASTPtr & ast)` and try to find the cache with data type name as "ast->range.first, ast->range.second" logging: Turn the logging level of "updateTableColumnInfo" into debug because that could cause lots of logging when restarting tiflash Signed-off-by: JaySon-Huang <tshent@qq.com> Co-authored-by: JaySon-Huang <tshent@qq.com> Co-authored-by: JaySon <tshent@qq.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com>

ti-chi-bot · 2025-07-14T06:17:32Z

In response to a cherrypick label: new pull request created to branch release-7.5: #10305.
But this PR has conflicts, please resolve them!

Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io>

ti-chi-bot · 2025-07-14T06:17:38Z

In response to a cherrypick label: new pull request created to branch release-8.1: #10306.
But this PR has conflicts, please resolve them!

Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io>

close #9947 Storage: Fix DataTypePtr is not shared as expected * Introduce a class `DataTypePtrCache` and manage the shared cache of `DataTypePtr` instances. * Introduce `DataTypeFactory::getOrSet(const ASTPtr & ast)` and try to find the cache with data type name as "ast->range.first, ast->range.second" logging: Turn the logging level of "updateTableColumnInfo" into debug because that could cause lots of logging when restarting tiflash Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io> Signed-off-by: JaySon-Huang <tshent@qq.com> Co-authored-by: JaySon <tshent@qq.com> Co-authored-by: JaySon-Huang <tshent@qq.com>

JaySon-Huang added 2 commits March 5, 2025 21:30

Fix DataTypePtr is not shared

d23f2a4

Signed-off-by: JaySon-Huang <tshent@qq.com>

Tune logging level

dcf5f32

Signed-off-by: JaySon-Huang <tshent@qq.com>

JaySon-Huang force-pushed the fix_shared_data_type branch from f183ef0 to dcf5f32 Compare March 5, 2025 13:30

ti-chi-bot bot removed the do-not-merge/needs-linked-issue label Mar 5, 2025

JaySon-Huang changed the title ~~[WIP] Storage: Fix DataTypePtr is not shared as expected~~ Storage: Fix DataTypePtr is not shared as expected Mar 5, 2025

JaySon-Huang requested review from CalvinNeo, JinheLin and Lloyd-Pottiger March 6, 2025 01:46

Add comment

c13d55f

Signed-off-by: JaySon-Huang <tshent@qq.com>

JaySon-Huang commented Mar 6, 2025

View reviewed changes

dbms/src/DataTypes/DataTypeFactory.h Outdated Show resolved Hide resolved

Apply suggestions from code review

2e96585

Lloyd-Pottiger approved these changes Mar 6, 2025

View reviewed changes

ti-chi-bot bot added needs-1-more-lgtm Indicates a PR needs 1 more LGTM. approved labels Mar 6, 2025

CalvinNeo reviewed Mar 6, 2025

View reviewed changes

JaySon-Huang added 3 commits March 6, 2025 12:12

Merge branch 'master' into fix_shared_data_type

7b8b309

shrink to reduce memory consumption

886961c

Signed-off-by: JaySon-Huang <tshent@qq.com>

Merge branch 'master' into fix_shared_data_type

299eacd

JinheLin approved these changes Mar 6, 2025

View reviewed changes

ti-chi-bot bot added lgtm and removed needs-1-more-lgtm Indicates a PR needs 1 more LGTM. labels Mar 6, 2025

ti-chi-bot bot added the needs-cherry-pick-release-8.5 Should cherry pick this PR to release-8.5 branch. label Mar 6, 2025

ti-chi-bot bot merged commit 24d3106 into pingcap:master Mar 6, 2025
5 checks passed

ti-chi-bot mentioned this pull request Mar 6, 2025

Storage: Fix DataTypePtr is not shared as expected (#9939) #9952

Merged

12 tasks

JaySon-Huang deleted the fix_shared_data_type branch March 6, 2025 09:04

ti-chi-bot bot removed the needs-cherry-pick-release-8.5 Should cherry pick this PR to release-8.5 branch. label Mar 27, 2025

ti-chi-bot bot added needs-cherry-pick-release-7.5 Should cherry pick this PR to release-7.5 branch. needs-cherry-pick-release-8.1 Should cherry pick this PR to release-8.1 branch. labels Jul 14, 2025

ti-chi-bot pushed a commit to ti-chi-bot/tiflash that referenced this pull request Jul 14, 2025

This is an automated cherry-pick of pingcap#9939

8d7938a

Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io>

ti-chi-bot mentioned this pull request Jul 14, 2025

Storage: Fix DataTypePtr is not shared as expected (#9939) #10305

Merged

12 tasks

ti-chi-bot pushed a commit to ti-chi-bot/tiflash that referenced this pull request Jul 14, 2025

This is an automated cherry-pick of pingcap#9939

bfe927a

Signed-off-by: ti-chi-bot <ti-community-prow-bot@tidb.io>

ti-chi-bot mentioned this pull request Jul 14, 2025

Storage: Fix DataTypePtr is not shared as expected (#9939) #10306

Merged

12 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Storage: Fix DataTypePtr is not shared as expected#9939

Storage: Fix DataTypePtr is not shared as expected#9939
ti-chi-bot[bot] merged 7 commits intopingcap:masterfrom
JaySon-Huang:fix_shared_data_type

JaySon-Huang commented Mar 4, 2025 •

edited

Loading

Uh oh!

Uh oh!

CalvinNeo left a comment

Uh oh!

ti-chi-bot bot commented Mar 6, 2025

Uh oh!

ti-chi-bot bot commented Mar 6, 2025

Uh oh!

Uh oh!

ti-chi-bot commented Mar 6, 2025

Uh oh!

ti-chi-bot commented Jul 14, 2025

Uh oh!

ti-chi-bot commented Jul 14, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

JaySon-Huang commented Mar 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What problem does this PR solve?

What is changed and how it works?

Check List

Release note

Uh oh!

Uh oh!

CalvinNeo left a comment

Choose a reason for hiding this comment

Uh oh!

ti-chi-bot bot commented Mar 6, 2025

Uh oh!

ti-chi-bot bot commented Mar 6, 2025

[LGTM Timeline notifier]

Uh oh!

Uh oh!

ti-chi-bot commented Mar 6, 2025

Uh oh!

ti-chi-bot commented Jul 14, 2025

Uh oh!

ti-chi-bot commented Jul 14, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

JaySon-Huang commented Mar 4, 2025 •

edited

Loading