auto">Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve. Reviewed July 11, 2026, 8:52 AM ET / 12:52 UTC.

Summary
Current main still uses contiguous serial packing for compact whole-test groups, while the direct candidate fix remains open at #104472. This protected, member-authored, assigned issue should remain open until that PR is reviewed and either merged or rejected.

Reproducibility: yes. at source level: current main deterministically places sorted whole groups into contiguous serial batches of six. The exact 403-second duration still requires the associated workflow timing artifact.

Root-cause cluster
Relationship: canonical
Canonical: #104471
Summary: This issue is the canonical report of the compact whole-job critical path, and the linked pull request is its direct candidate fix.

Members:

fixed_by_candidate: improve(ci): balance compact whole test jobs #104472 - The open pull request directly changes the reported packing algorithm and adds focused regression coverage.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Next step
The active paired PR already owns the bounded implementation, so creating another repair job would duplicate current maintainer work.

Review details

Best possible solution:

Review and land the paired PR if exact-head CI timings confirm that it lowers the slowest compact whole job while preserving runner/dist partitions, job count, and runner-registration budget.

Do we have a high-confidence way to reproduce the issue?

Yes at source level: current main deterministically places sorted whole groups into contiguous serial batches of six. The exact 403-second duration still requires the associated workflow timing artifact.

Is this the best way to solve the issue?

Yes in direction: balancing the existing groups within the same runner/dist partitions and job budget is narrower and safer than increasing fanout, runner caps, or timeouts.

AGENTS.md: found and applied where relevant.

Remaining risk / open question:

The source establishes the serial packing mechanism, but the issue and visible discussion do not provide the workflow artifact behind the exact 403-second measurement or after-change timing proof.

Codex review notes: model internal, reasoning high; reviewed against 3f33f0bb5f3e.

Label changes

Label justifications:

P2: This is a concrete CI latency problem with a narrow active repair path and no shipped-product outage.
impact:other: The reported long critical path materially delays maintainer CI feedback outside the specific product-impact categories.

Evidence reviewed

What I checked:

Protected live issue: The issue is open, authored and assigned to repository member vincentkoc, and carries the maintainer label, requiring explicit maintainer handling rather than automatic closure.
Current-main root cause: Within each runner and requiresDist partition, current main sorts whole groups and slices them into contiguous batches of six, allowing expensive neighboring groups to share one serial job. (scripts/lib/ci-node-test-plan.mjs:1128, 3f33f0bb5f3e)
Active CI entry point: The main CI workflow imports the Node test planner and requests its compact plan, making the reported packing behavior active on current main. (.github/workflows/ci.yml:303, 3f33f0bb5f3e)
Direct candidate fix: The paired PR preserves the computed whole-job count while distributing groups round-robin and adds focused assertions for balanced group counts and separation of the named heavy groups. (scripts/lib/ci-node-test-plan.mjs:1128, 59ef0cf651c6)
Candidate remains active: The paired PR remains open and GitHub reports it mergeable, making it an implementation candidate rather than evidence that this issue is already fixed. (59ef0cf651c6)
Feature provenance: Blame and symbol history tie the compact bundling path on current main to commit 0af8467 by Peter Steinberger, supporting routing alongside the current issue owner. (scripts/lib/ci-node-test-plan.mjs:1007, 0af846713749)

Likely related people:

vincentkoc: The issue is assigned to vincentkoc, who authored the active planner and regression-test changes and previously contributed release CI work. (role: current issue owner and recent CI contributor; confidence: high; commits: e085fa1a3ffd, 18013b91ee6f, 59ef0cf651c6; files: scripts/lib/ci-node-test-plan.mjs, test/scripts/ci-node-test-plan.test.ts)
Peter Steinberger: Git blame and symbol history attribute the current compact bundling implementation and its CI integration to commit 0af8467. (role: introduced current compact planner behavior; confidence: high; commits: 0af846713749; files: scripts/lib/ci-node-test-plan.mjs, .github/workflows/ci.yml, test/scripts/ci-node-test-plan.test.ts)

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

Uh oh!

Uh oh!

CI compact whole shard packing creates a 403s critical path #104471

Description

Metadata

Metadata

Assignees

Labels

Type

Fields

Projects

Milestone

Relationships

Development

Issue actions