session: make TxnInfo() return even if process info is empty

[tangenta]

What problem does this PR solve?

Issue Number: close #57043

Problem Summary:

tidb/pkg/session/session.go

Lines 4026 to 4030 in c240fdf

	txnInfo := tmp.TxnInfo()
	if txnInfo != nil {
	startTS = txnInfo.StartTS
	processInfoID = txnInfo.ConnectionID
	}

TxnInfo() could be nil if process info is not set(see the implementation of TxnInfo()). Such sessions are commonly used in DDL/DXF:

tidb/pkg/ddl/backfilling_operators.go

Lines 533 to 539 in c240fdf

	sessCtx, err := w.sessPool.Get()
	if err != nil {
	logutil.Logger(w.ctx).Error("tableScanWorker get session from pool failed", zap.Error(err))
	w.ctx.onError(err)
	return
	}
	w.se = session.NewSession(sessCtx)

And they only invoke session.Begin() before using:

tidb/pkg/ddl/index_cop.go

Lines 43 to 48 in c240fdf

	func wrapInBeginRollback(se *sess.Session, f func(startTS uint64) error) error {
	err := se.Begin(context.Background())
	if err != nil {
	return errors.Trace(err)
	}
	defer se.Rollback()

As a result, the startTS from internal DDL sessions will not be reported to etcd, and GC safepoint will be advanced unexpectedly.

What changed and how does it work?

Make TxnInfo() return even if process info is empty.

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 @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

Attention: Patch coverage is 89.79592% with 5 lines in your changes missing coverage. Please review.

Project coverage is 57.8279%. Comparing base (1de6f3e) to head (0f0569e).
Report is 15 commits behind head on master.

Additional details and impacted files

@@                Coverage Diff                @@
##             master     #57044         +/-   ##
=================================================
- Coverage   73.1785%   57.8279%   -15.3506%     
=================================================
  Files          1655       1808        +153     
  Lines        456448     653865     +197417     
=================================================
+ Hits         334022     378117      +44095     
- Misses       101907     250558     +148651     
- Partials      20519      25190       +4671     

Flag	Coverage Δ
integration	40.2998% <51.0204%> (?)
unit	73.1293% <89.7959%> (+0.6713%)	⬆️

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

Components	Coverage Δ
dumpling	52.9478% <ø> (ø)
parser	∅ <ø> (∅)
br	62.8718% <ø> (+17.1715%)	⬆️

[YangKeao]

sessionVars.TxnCtx will not be cleaned before executing another statement, so using the StartTS from TxnCtx will block the GC for very long time if some internal sessions are not used (e.g. stored in a pool).

It's possible to clean the sessionVars.TxnCtx. But IMO, it'd be simpler to use:

txn := tmp.Txn(false)
if txn != nil {
    startTS = txn.StartTS()
}

It should have the same value with txnInfo.StartTS (ref changePendingToValid, it sets the txnInfo.StartTS with t.StartTS()), and it doesn't depend on the ProcessInfo.

[tangenta]

2cfe636

[tangenta]

Maybe we still need TxnInfo() because processInfoID will be used to filter analyze SQL.

[tangenta]

/retest

[tiprow]

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

In response to this:

/retest

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.

[YangKeao]

LGTM

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2024-11-05 03:44:44.61158894 +0000 UTC m=+926197.450744486: ☑️ agreed by YangKeao.
• 2024-11-05 04:04:02.988100501 +0000 UTC m=+927355.827256056: ☑️ agreed by lcwangchao.

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: D3Hunter, lcwangchao, YangKeao

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,YangKeao,lcwangchao]
• pkg/ddl/OWNERS [D3Hunter]
• pkg/domain/OWNERS [D3Hunter]
• pkg/infoschema/OWNERS [D3Hunter]

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

[tangenta]

/retest

[tiprow]

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

In response to this:

/retest

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.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-8.5: #57161.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-8.1: #57163.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.1: #57162.

[ti-chi-bot]

In response to a cherrypick label: new pull request created to branch release-7.5: #57164.