Repository navigation
ci(lanes): split two long test files and run console browser 2 on 32vcpu - #2307
Conversation
…es and run console browser 2 on 32vcpu Console Browser 2 was CPU-bound on 16vcpu (4 CPUs): its files' summed time tracked the lane's. Move it to 32vcpu (8 CPUs) and move the provisioning-retry, model-selection, conflict and duplicate-write cases from console-agents.test.mjs into console-agents-2.test.mjs unchanged. PostgreSQL Application's longest file was revision-teardown. Move its Agent stop, drain, deletion and Namespace teardown cases unchanged into postgres-worker-agent-revision-stop.test.mjs; repository session cases stay. Repository Credentials Platform also runs on 32vcpu as a measured trial.
|
🦞👀 Pull request received. I will update this pull request when review starts. ClawSweeper review completeClawSweeper finished reviewing this revision. The review result is being finalized. |
GitHub Actions
Error
GitHub Actions
Error
clawsweeper
Bot
added
P3
…lit siblings by suffix On 32vcpu the Repository Credentials Platform lane took as long: its runtime image import was 35 s (28 s on 16vcpu) and its long case 82 s (81-91 s), so it is not CPU-bound and stays on 16vcpu. The new files' comments named the file they were split from. CI Impact runs full CI for a test file that another file names, so name siblings by suffix. List Console Browser 2 among the 32vcpu lanes in the CI docs.
|
Codex review: needs maintainer review before merge. What this changesThis PR splits two long test files, updates their lane inventories, and moves Console Browser 2 to a larger runner. Example: Console Browser 2 declares fileConcurrency 8.
Review scores
ProductKind: Performance · Worth it: Yes · Fix scope: Complete Merge readiness✅ Ready for maintainer review Keep open for maintainer triage: the focused CI optimization preserves test coverage, resolves the prior documentation finding, and introduces no actionable defect found in this review. Priority: P3 Before mergeNone. FindingsNone. Agent review detailsHow this fits togetherGitHub Actions selects runners and lane manifests; the CI scheduler executes isolated browser and PostgreSQL fixtures and publishes results consumed by CI Required. flowchart TB
A[GitHub Actions] --> B[Runner selection]
A --> C[Lane manifests]
B --> D[CI scheduler]
C --> D
D --> E[Isolated test fixtures]
E --> F[CI Required]
Technical reviewBest possible solution: Retain the coverage-preserving split and measured browser runner increase through the existing scheduler. Do we have a high-confidence way to reproduce the issue? Source confirms the CPU-based concurrency limit and serial file organization; the reported performance bottleneck was not independently benchmarked during this read-only review. Is this the best way to solve the issue? The change extends established scheduling and fixture boundaries while preserving assertions, required names, and isolation. AGENTS.md: found and applied where relevant. Codex review notes: model internal, reasoning medium; reviewed against 3d180654c2d7. Provenance checked
TestingProof path: in-process harness. SecurityNone. EvidenceWhat I checked:
Likely related people:
Review metrics
LabelsLabel changes:
Label justifications:
Rating scale6/6 🦀 challenger crab · 5/6 🦞 diamond lobster · 4/6 🐚 platinum hermit · 3/6 🦐 gold shrimp · 2/6 🦪 silver shellfish · 1/6 🧂 unranked krab. Overall follows the weaker of proof and patch quality; ✨ marks media proof (a screenshot, video, or linked artifact) that directly shows the changed behavior. WorkflowClawSweeper edits this one comment on every review. Comment HistoryReview history (1 earlier review cycle)
Reviewed October 13, 2026, 7:12 AM ET / 11:12 UTC (Revision 2). |
GitHub Actions
Failure
GitHub Actions
Failure
clawsweeper
Bot
added
status: 👀 ready for maintainer look
Problem
After #2303, CI Required ends about +203..+270 s and three lanes finish last, nearly tied:
fileConcurrency: 8but ran four files at a time, and its ten files' summed time (409-596 s) tracked its test window (112-159 s).console-agents.test.mjsruns its 46 cases one after another for 112-159 s; locally, uncontended, the same file takes 66 s.postgres-worker-agent-revision-teardown.test.mjs(132-180 s, 61 cases), not the revision file (109-153 s).Change
blacksmith-32vcpu-ubuntu-2404(8 CPUs), inci.ymland in full-integration's same-SHA runner expression.test()calls, about 30 % of the file's time) move unchanged fromconsole-agents.test.mjstoconsole-agents-2.test.mjs. The moved block avoids the lines that open PRs feat: assign configured OpenClaw roles to Agent users #906, Enable Codex OAuth with OpenShell-owned credentials #1559, docs(specs): propose rfc-0019 for repository options without a repository driver #878, Enable dedicated OpenClaw from Console with pinned native runtime #1685 and feat: add Agent workload tags #115 change.postgres-worker-agent-revision-stop.test.mjs, with thecoldSshComputeDriverhelper that only they use. The repository session cases stay.expectedTestssplit to match.parallelFilesin both manifests are ordered longest first, using measured averages.ci.ymlcomment records this.docs/flows/controller-worker.mdname the new file, anddocs/testing/ci.mdanddocs/flows/github-actions-testing.mdlist Console Browser 2 as a 32vcpu lane. Word counts are unchanged: 2493, 2485 and 2495.No required check name,
needs, or impact-gate logic changes.Verification
Moved cases are byte-identical to their originals (diffed after the split); only unused imports were removed. Audit:
run-tests.mjs audithas no issues.actionlintis clean, and so are eslint and prettier on the touched files.Local:
ci-impact.test.mjs64/64;console-agents.test.mjs31/31 (40 s);console-agents-2.test.mjs15/15 (26 s).Open-PR conflicts (
git merge-treeagainst main vs this head): only fix(worker): settle initial provisioning on a superseding deploy or stop #2268 (ours) gains one conflict hunk, where its new case lands in the moved block. fix(worker): keep an active runtime when repository authority fails #2237, test: mint isolated QA observer tokens from the GitHub App #2106, fix(ci): synchronize Agent watches before starting test files #2012, fix(ci): retain nested failure causes in diagnostics #2002, Enable dedicated OpenClaw from Console with pinned native runtime #1685, Enable Codex OAuth with OpenShell-owned credentials #1559, feat: assign configured OpenClaw roles to Agent users #906, docs(specs): propose rfc-0019 for repository options without a repository driver #878 and feat: add Agent workload tags #115 are unchanged.PR CI: CI Required green on both heads (runs 38387511347 on 34b2abd and 38388164698 on fd994d9). Main runs are from 10-13 09:03-10:41Z (n=6).
console-agentsfile(s)CI Required's end barely moves yet: the k3d fixture lanes, Image Runtime Startup and PostgreSQL Application now end within about 15 s of one another (+191..+207 in these runs). The lane cuts remove Console Browser 2 from that group (it reached +228 on main) and narrow PostgreSQL Application's range.
Risks