logpuller: optimize resolved-ts dispatch allocations

[asddongmen]

What problem does this PR solve?

Issue Number: close #4697

What is changed and how it works?

• Added a single-region fast path:
  • If len(resolvedTsEvent.Regions) == 1, directly push one-state event.
• Changed initial resolvedStates capacity to:
  • capHint = min(len(resolvedTsEvent.Regions), resolvedTsStateBatchSize)
• Kept large-batch protection:
  • resolvedTsStateBatchSize remains 1024.
• Removed redundant allocation pattern:
  • After flush, do not always allocate a new slice.
  • Allocate the next batch buffer only when there are remaining regions to process.
• Added tests and benchmarks in region_request_worker_test.go:
  • Single-region fast-path behavior
  • Large-batch split behavior
  • Legacy vs current benchmark comparisons

Benchmark Command

mkdir -p /tmp/go-cache /tmp/go-tmp && \
GOCACHE=/tmp/go-cache GOTMPDIR=/tmp/go-tmp \
go test ./logservice/logpuller \
  -run '^$' \
  -bench 'BenchmarkDispatchResolvedTsEvent' \
  -benchmem \
  -count=5

Average Comparison (5 runs)

Scenario	Legacy ns/op	Current ns/op	ns/op Change	Speedup	Legacy B/op	Current B/op	B/op Change	Legacy allocs/op	Current allocs/op	allocs Change
SingleRegion	664.94	30.19	-95.46%	22.02x	19074	137	-99.28%	2.00	1.00	-50.00%
SmallBatch	658.94	148.20	-77.51%	4.45x	19071	256	-98.66%	2.00	1.00	-50.00%
LargeBatch	36325.80	36532.80	+0.57%	0.99x	47899	38424	-19.78%	5.00	4.00	-20.00%

• SingleRegion and SmallBatch show major CPU and allocation improvements.
• LargeBatch keeps bounded memory behavior and reduces allocations; ns/op is roughly flat (+0.57%).

Check List

Tests

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

Questions

Will it cause performance regression or break compatibility?

Do you need to update user documentation, design documentation or monitoring documentation?

Release note

  Reduce CPU and allocation overhead in log puller by optimizing resolved-ts event dispatch for small batches.

Summary by CodeRabbit

• Performance

  • Improved resolved-timestamp dispatch: smarter pre-sizing, buffer reuse, and adaptive batching for single-region fast path and large multi-region splits, reducing allocations and improving throughput.

• Tests

  • Added unit tests and benchmarks validating single-region fast-path, large-set batch-splitting, and performance comparisons against the legacy dispatcher.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 4e756c24-8e9c-4105-b60c-46467824015a

📥 Commits

Reviewing files that changed from the base of the PR and between 3383cc7 and f83f247.

📒 Files selected for processing (2)

• logservice/logpuller/region_request_worker.go
• logservice/logpuller/region_request_worker_test.go

🚧 Files skipped from review as they are similar to previous changes (2)

• logservice/logpuller/region_request_worker_test.go
• logservice/logpuller/region_request_worker.go

---

📝 Walkthrough

Walkthrough

Compute resolvedStates capacity from incoming region count (capped by batch size), use index-based iteration to track remaining regions, avoid unnecessary reallocation by nil-ing the slice after pushing. Add tests and benchmarks validating single-region fast-path and large-batch splitting.

Changes

Cohort / File(s)	Summary
Dispatch logic logservice/logpuller/region_request_worker.go	Initialize resolvedStates capacity from len(resolvedTsEvent.Regions) (capped by resolvedTsStateBatchSize); switch to index-based loop to compute remaining regions; pass slice directly to push, set resolvedStates = nil after flush, and only reallocate when remaining elements exist.
Tests & benchmarks logservice/logpuller/region_request_worker_test.go	Add mockRegionEventDynamicStream, helper to build test worker and synthetic ResolvedTs events, unit tests asserting single-region push and 2050 -> 1024/1024/2 splitting, and benchmarks comparing legacy fixed-1024 batching vs new dispatch logic for various region counts.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• logpuller: fix region leak after remove subscription #4267 — Modifies dispatchResolvedTsEvent batching/flush behavior in the same file.

Suggested labels

size/M

Suggested reviewers

• wk989898
• lidezhu
• flowbehappy

Poem

🐇 I hopped through slices, small and fleet,
Sized each batch to fit the beat.
Nil the tray, reuse what's near,
Fewer hops, the path is clear.
Pushes counted — merry cheer!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely summarizes the main change: optimizing memory allocations in the resolved-ts dispatch logic within the log puller.
Description check	✅ Passed	The description includes issue number (close #4697), detailed explanation of changes, benchmark results, test additions, and a release note—fully complying with the template requirements.
Linked Issues check	✅ Passed	All coding requirements from issue #4697 are met: single-region fast path, sized initial capacity, large-batch protection, reduced redundant allocations, and comprehensive tests/benchmarks validating the optimization.
Out of Scope Changes check	✅ Passed	All changes are strictly scoped to resolving the allocation optimization issue; no unrelated code modifications are present outside the linked issue objectives.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

• X
• Mastodon
• Reddit
• LinkedIn

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

[gemini-code-assist]

Code Review

This pull request optimizes the dispatchResolvedTsEvent function by introducing a fast path for single-region events and refining memory allocation during batch processing of multi-region events. Review feedback suggests downgrading a new warning log to debug level to prevent potential log noise and further optimizing slice re-allocations by calculating a more precise capacity hint for remaining regions.

[gemini-code-assist]

This log.Warn is new and could be noisy if many single-region ResolvedTs events are received for untracked regions. Consider changing it to log.Debug or removing it, as untracked regions are common during subscription changes.

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hongyunyan, lidezhu

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 [hongyunyan,lidezhu]

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:

• 2026-04-04 05:50:59.857827585 +0000 UTC m=+589865.063187642: ☑️ agreed by hongyunyan.
• 2026-04-04 11:17:58.290557924 +0000 UTC m=+609483.495917981: ☑️ agreed by lidezhu.

[asddongmen]

/test all

[asddongmen]

/test all