domain,infoschema: make infoschema activity block GC safepoint advancing

[tiancaiamao]

What problem does this PR solve?

Issue Number: close #57952

Problem Summary:

What changed and how does it work?

Currently, when we get a infoschema v2 instance, it's lifetime is in range [ts, ts + 10min)
If the caller hold a infoschema, and use it some time later, it might get error "GC lifetime is shorter than transaction duration" error. That's because infoschema v2 internally use meta package API, while the meta API does not block the GC safepoint advancing.

In this commit, I add a keepAlive() call to the infoschema API,
that keepAlive() function will keep the minimal start ts of active infoschema API calls.

The infosync.Syncer call ReportMinStartTS() periodically, amd now it will take the active infoschema ts into consideration, which is updated by the keepAlive() function.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• 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 @tiancaiamao. 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 75.0849%. Comparing base (bb9096c) to head (aef6ee9).
Report is 10 commits behind head on master.

Additional details and impacted files

@@               Coverage Diff                @@
##             master     #58062        +/-   ##
================================================
+ Coverage   73.1711%   75.0849%   +1.9138%     
================================================
  Files          1674       1720        +46     
  Lines        461507     472153     +10646     
================================================
+ Hits         337690     354516     +16826     
+ Misses       103079      95413      -7666     
- Partials      20738      22224      +1486     

Flag	Coverage Δ
integration	49.0231% <100.0000%> (?)
unit	72.3040% <100.0000%> (+0.0068%)	⬆️

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

Components	Coverage Δ
dumpling	52.6910% <ø> (ø)
parser	∅ <ø> (∅)
br	61.5989% <ø> (+15.6156%)	⬆️

[tiancaiamao]

One possible case that this fix can not cover is that, the caller take an infoschema V2 instance but never use it.
Then keepAlive() is not called.
10min later, infoschema v2 API is called on this is for the first time...

This case is rare and if we found it we can adjust at the caller side, so it should not be a real issue.

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2024-12-09 02:59:50.936194742 +0000 UTC m=+234581.024997280: ☑️ agreed by wjhuang2016.
• 2024-12-09 05:02:50.825493348 +0000 UTC m=+241960.914295890: ☑️ agreed by lcwangchao.

[tiancaiamao]

/hold

[MyonKeminta]

Can we use atomic.Uint64 instead?

[MyonKeminta]

Is it possible that an infoschemaV2 is acccessed in a lower frequence than ReportMinStartTS, so that a tick of ReportMinStartTS may miss this infoschemaV2 instance, which allows GC to proceed, and then the infoschemaV2 instance becomes unavailable again?

[tiancaiamao]

In theory, that is possible, so there is still potential risk. I have discussed that with @lcwangchao
But in practice I think this fix is good enough for the most common cases. @MyonKeminta

[MyonKeminta]

Ok. But I suggest these details can be noted in the comments in the code.
Rest LGTM.

[tiancaiamao]

Done

[tiancaiamao]

@MyonKeminta we need some review from the transaction team, PTAL

[tiprow]

@tiancaiamao: The following test failed, say /retest to rerun all failed tests or /retest-required to rerun all mandatory failed tests:

Test name	Commit	Details	Required	Rerun command
fast_test_tiprow	9f75e35	link	true	/test fast_test_tiprow

Full PR test history. Your PR dashboard.

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. I understand the commands that are listed here.

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: lcwangchao, MyonKeminta, 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 [MyonKeminta,lcwangchao,wjhuang2016]
• pkg/ddl/OWNERS [wjhuang2016]
• pkg/domain/OWNERS [wjhuang2016]
• pkg/infoschema/OWNERS [wjhuang2016]

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

[tiancaiamao]

/test pull-unit-test-ddlv1

[tiprow]

@tiancaiamao: Cannot trigger testing until a trusted user reviews the PR and leaves an /ok-to-test message.

In response to this:

/test pull-unit-test-ddlv1

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.

[tiancaiamao]

/unhold

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-8.5: #58103.