storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock by lcwangchao · Pull Request #19374 · tikv/tikv

lcwangchao · 2026-02-14T04:48:04Z

What is changed and how it works?

Issue Number: Close #19312, Close #19393

What's Changed:

storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock

Problem Summary:
- When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts.

What changed:
- Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version.
- Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions.

Tests:
- Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`.
- Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`.
- Verify existing related tests:
  - `test_point_get_load_commit_ts`
  - `test_scan_with_load_commit_ts`

Related changes

PR to update pingcap/docs/pingcap/docs-cn:
Need to cherry-pick to the release branch

Check List

Tests

Unit test
Integration test
Manual test (add detailed scripts or steps below)
No code

Side effects

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

Release note

Fix wrong commit_ts returned by Get and TableScan when `load_commit_ts` is enabled and newer `Op_Lock` versions exist above the latest `Op_Put`.

coderabbitai · 2026-02-14T04:48:21Z

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

🔍 Trigger review

✨ Finishing touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Post copyable unit tests in a comment

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

Connor1996 · 2026-02-14T05:22:13Z

src/storage/mvcc/reader/scanner/backward.rs

-            }
            if current_ts <= last_checked_commit_ts {
                // We reach the last handled key
                return self.handle_last_version(last_version, user_key, loaded_commit_ts);


What if it hits this branch, the loaded_commit_ts is not updated

Good question. In this branch we return last_version (captured during the initial prev() loop), not the write at current_ts. loaded_commit_ts is updated together with last_version (only for PUT/DELETE), so it already matches what we return here. Not updating it for LOCK/ROLLBACK is intentional to avoid the original bug where commit_ts could be taken from a newer LOCK. Added test_load_commit_ts_with_top_lock_versions to cover the case (many committed LOCKs on top of a visible PUT) and it asserts the returned commit_ts comes from the PUT.

lcwangchao · 2026-02-14T05:46:47Z

/retest

Connor1996

LGTM

Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

ti-chi-bot · 2026-02-14T06:02:57Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: cfzjywxk, Connor1996

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~~ [Connor1996,cfzjywxk]

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 · 2026-02-14T06:02:58Z

[LGTM Timeline notifier]

Timeline:

2026-02-14 05:49:45.768687408 +0000 UTC m=+74753.515081436: ☑️ agreed by Connor1996.
2026-02-14 06:02:57.622292298 +0000 UTC m=+75545.368686331: ☑️ agreed by cfzjywxk.

…_Lock (tikv#19374) close tikv#19312 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

…_Lock (#19374) (#19394) close #19393 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

…_Lock (tikv#19374) close tikv#19312 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

…_Lock (#19374) (#19394) (#19397) close #19393 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com> Co-authored-by: Chao Wang <cclcwangchao@hotmail.com>

…_Lock (tikv#19374) (tikv#19394) close tikv#19393 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

…_Lock (tikv#19374) (tikv#19394) close tikv#19393 storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock Problem Summary: - When `load_commit_ts=true` and a key has newer committed `Op_Lock` versions on top of the visible `Op_Put`, TiKV may return the lock commit_ts instead of the put commit_ts. What changed: - Fix `PointGetter` to return the commit_ts of the actual write version loaded by `last_change` shortcut, instead of the current cursor version. - Fix backward scanner to only update `loaded_commit_ts` when selecting a `Put` version, avoiding ts overwrite from `Lock/Rollback` versions. Tests: - Add `test_point_get_load_commit_ts_with_top_lock_versions` in `point_getter.rs`. - Add `test_load_commit_ts_with_top_lock_versions` in `scanner/backward.rs`. - Verify existing related tests: - `test_point_get_load_commit_ts` - `test_scan_with_load_commit_ts` Signed-off-by: Chao Wang <cclcwangchao@hotmail.com> Signed-off-by: rishabh mittal <mittalrishabh@gmail.com>

ti-chi-bot bot added release-note Denotes a PR that will be considered when it comes time to generate release notes. dco-signoff: yes Indicates the PR's author has signed the dco. size/M Denotes a PR that changes 30-99 lines, ignoring generated files. labels Feb 14, 2026

mvcc: wrong commit_ts in Get/TableScan when latest write is Op_Lock

a76b33f

Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

lcwangchao force-pushed the fix_commit_ts_with_lock branch from b6953d6 to a76b33f Compare February 14, 2026 04:56

fix lint

214bf9c

Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

Connor1996 reviewed Feb 14, 2026

View reviewed changes

Connor1996 approved these changes Feb 14, 2026

View reviewed changes

ti-chi-bot bot added needs-1-more-lgtm Indicates a PR needs 1 more LGTM. approved labels Feb 14, 2026

add more tests

47acbc9

Signed-off-by: Chao Wang <cclcwangchao@hotmail.com>

ti-chi-bot bot added size/XL Denotes a PR that changes 500-999 lines, ignoring generated files. and removed size/M Denotes a PR that changes 30-99 lines, ignoring generated files. labels Feb 14, 2026

cfzjywxk approved these changes Feb 14, 2026

View reviewed changes

ti-chi-bot bot added lgtm and removed needs-1-more-lgtm Indicates a PR needs 1 more LGTM. labels Feb 14, 2026

ti-chi-bot bot merged commit 055c58f into tikv:release-8.5-20260121-v8.5.5 Feb 14, 2026
5 checks passed

lcwangchao deleted the fix_commit_ts_with_lock branch February 14, 2026 06:34

lcwangchao mentioned this pull request Feb 25, 2026

The TiKV may return wrong _tidb_commit_ts for point-get and reverse-scan #19393

Closed

Copilot AI mentioned this pull request Feb 25, 2026

storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock (#19374) #19394

Merged

9 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock#19374

storage: fix wrong commit_ts in Get/TableScan when latest write is Op_Lock#19374
ti-chi-bot[bot] merged 3 commits intotikv:release-8.5-20260121-v8.5.5from
lcwangchao:fix_commit_ts_with_lock

lcwangchao commented Feb 14, 2026 •

edited

Loading

Uh oh!

coderabbitai bot commented Feb 14, 2026 •

edited

Loading

Review skipped

Uh oh!

Connor1996 Feb 14, 2026

Uh oh!

lcwangchao Feb 14, 2026

Uh oh!

lcwangchao commented Feb 14, 2026

Uh oh!

Connor1996 left a comment

Uh oh!

ti-chi-bot bot commented Feb 14, 2026

Uh oh!

ti-chi-bot bot commented Feb 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

lcwangchao commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What is changed and how it works?

Related changes

Check List

Release note

Uh oh!

coderabbitai bot commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review skipped

Uh oh!

Connor1996 Feb 14, 2026

Choose a reason for hiding this comment

Uh oh!

lcwangchao Feb 14, 2026

Choose a reason for hiding this comment

Uh oh!

lcwangchao commented Feb 14, 2026

Uh oh!

Connor1996 left a comment

Choose a reason for hiding this comment

Uh oh!

ti-chi-bot bot commented Feb 14, 2026

Uh oh!

ti-chi-bot bot commented Feb 14, 2026

[LGTM Timeline notifier]

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

lcwangchao commented Feb 14, 2026 •

edited

Loading

coderabbitai bot commented Feb 14, 2026 •

edited

Loading