Skip to content

ci: finish the m1 workflow signal and cost audit - #6266

Merged
kixelated merged 15 commits into
mainfrom
quest/m1/audit-ci
Oct 22, 2026
Merged

kixelated merged 15 commits into
mainfrom
quest/m1/audit-ci

Conversation

@kixelated

@kixelated kixelated commented Oct 22, 2026 •

Copy link
Copy Markdown
Collaborator

Problem

The m1 CI workflow audit found Nightly red every night, with existing measurements visible only in logs or artifacts. A red Nightly hides the next regression.

Approach

  • Inventory all 49 workflows and the latest scheduled Nightly's jobs with signal, visibility, and cost verdicts. Classify every failed job of the 2026-10-21 and 2026-10-22 Nightly, Interop, and TS runs against what merged since.
  • Publish the existing generator, Rust and JS coverage, browser baseline, and browser benchmark measurements to Actions job summaries.
  • Gate the Platform, x11, and size uploads on their producing step. On 2026-10-22 the macOS job crashed in the Swift check, then reported the skipped benchmark's "No files were found" as its first error.
  • Register js/hang/bench/jitter.ts (moq-dev/moq#6359) and js/net/bench/subscribe-tracks.ts (moq-dev/moq#6336) in Nightly. Without them, just gh test fails on main and Nightly's gh test job would go red tonight.
  • Replay the historical failures on current main. The moq-mux and moq-net feature lints pass. Loom still fails. moq-dev/moq#6279 fixed the connection fuzz crash, using the same 847-byte input.
  • Finish this run: delete quest/m1/audit-ci.md and add quest/m2/audit-ci.md at the end of m2's Required, gated on the release cut. It carries over the per-workflow table, the baselines, and what to check first.

Main superseded this PR's setup and feature fixture fixes (moq-dev/moq#6325) and the archive producer benchmark (moq-dev/moq#6326), so the merge dropped them.

Impact

  • Public API and wire: none.
  • Nightly, Platform, and x11 keep their schedules, assertions, and artifacts. Job summaries gain tables. Three uploads now skip when their producing step never ran. Nightly's JS group runs two more benches.

Validation

  • just check and quest check --base origin/main pass on the head. just gh test passes, and fails on main without the bench registration.
  • just js jitter-bench and just js bench subscribe-tracks run to completion.
  • just rs net-features and cargo clippy -p moq-mux --all-targets --all-features -- -D warnings pass on main f42b336f0e.
  • Loom on main f42b336f0e: a_front_sees_a_reader_return exceeds Loom's branch limit, reproduced locally.
  • just test wasm on main f42b336f0e fails ietf: reads a published broadcast on two runs. A first-parent bisect points to moq-dev/moq#6348.
  • No Nightly was dispatched on this branch. The workflow edits are checked by actionlint, and none of them can run before the next scheduled night.

Follow-ups

These Nightly failures have no owner. Each is confirmed on current main or on the latest night, and is recorded in the m2 quest's "Check first":

  1. Loom: a_front_sees_a_reader_return exceeds the branch limit. The two a_*_reader_sees_the_front_end_a_refused_track models also run until the job's 60-minute timeout.
  2. Windows only: reconnect::a_protocol_violation_ends_reconnect (added by moq-dev/moq#6162) redials after a protocol violation over TCP.
  3. WASM: the IETF read case started failing with moq-dev/moq#6348. Join after a quick finish owns the behavior, but it sits in m2 while it now breaks Nightly on main.

Also: Tooling tests gate every merge is unclaimed. It would have caught the bench registration and setup fixture breaks in their own PRs.

(Written by Claude Opus 5.5)

🤖 Generated with Claude Code

kixelated and others added 7 commits October 21, 2026 22:49
Quest-Claim: bdbd54bd0bf7b32e
Co-Authored-By: GPT-6 <noreply@openai.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Preserve the exact current-main QUIC reproducer for the maintainer's scope decision without changing transmit behavior or starting a follow-up quest.

Co-Authored-By: GPT-6 <noreply@openai.com>
@kixelated kixelated changed the title CI workflow audit ci: checkpoint workflow signal and cost audit Oct 22, 2026
@kixelated

Copy link
Copy Markdown
Collaborator Author

Audit checkpoint pushed at c033627. The quest contains all 49 workflow and 166 Nightly job rows, failure classifications, measured repository billing totals, explicit timing-data limits, and the exact connection crash bytes plus reproduction command.

Full just ci passed before the clean rebase onto main bce217e. The rebased admission gate never admitted; its owned waiter has been cancelled, so full current-head validation is pending. No admission bypass was used. Quest, Markdown and whitespace checks pass after rebase.

This remains a draft. The maintainer choice is whether to scope the confirmed transmit crash as an independently reviewed m0 fix (recommended) or include it here. No protocol source change or follow-up quest has been started. Existing coverage and caching decisions are preserved. No public API or wire impact.

(Written by GPT-6)

kixelated and others added 7 commits October 22, 2026 15:26
# Conflicts:
#	sh/gh/nightly.sh
#	sh/gh/setup.test.ts
The 2026-10-22 Nightly's macOS Platform job crashed in the Swift check, then
reported "No files were found" from the skipped benchmark's upload as its first
error. Gate each `!cancelled()` upload that requires files on its producing
step, so an earlier failure stays the job's only error.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Classify the 2026-10-22 Nightly, Interop, and TS failures, replay the
historical Loom and feature-lint failures on current main, and carry the
per-workflow table, baselines, and what to check first into the m2 run, gated
on the release cut.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#6359 added js/hang/bench/jitter.ts and #6336 js/net/bench/subscribe-tracks.ts
without registering them, so `just gh test` fails on main and Nightly's
`gh test` would go red on its next run.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@kixelated kixelated changed the title ci: checkpoint workflow signal and cost audit ci: finish the m1 workflow signal and cost audit Oct 22, 2026
@kixelated
kixelated marked this pull request as ready for review October 22, 2026 22:57
@kixelated

Copy link
Copy Markdown
Collaborator Author

@codex review

(Written by Claude Opus 5.5)

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-22T23:03:07.186978Z — Manual request
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@coderabbitai

coderabbitai Bot commented Oct 22, 2026

Copy link
Copy Markdown
Contributor

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 24 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used all 4 included reviews currently available.

Learn how review limits work.

Review configuration:

⚙️ Run configuration
  • Source mode: Auto
  • Repository: Unavailable
  • Branch: Unavailable
  • File: No configuration file from this repository applied
  • Commit: Unavailable
  • Layers: Organization UI
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: ea0d729c-ceb2-4aa4-8c7e-0456cf91b677
📥 Commits

Reviewing files that changed from the base of the PR and between 6631674 and c4a45d0.

📒 Files selected for processing: 19
  • .github/workflows/nightly.yml
  • .github/workflows/platform.yml
  • .github/workflows/x11.yml
  • bench/browser/driver.ts
  • bench/ffi-gen/README.md
  • bench/ffi-gen/report.py
  • js/justfile
  • quest/m0/ci/queue-allgreen.md
  • quest/m1/README.md
  • quest/m1/audit-ci.md
  • quest/m1/audit-compile-time.md
  • quest/m1/ci-housekeeping.md
  • quest/m2/README.md
  • quest/m2/audit-ci.md
  • quest/m2/audit-test-cost.md
  • sh/gh/nightly.sh
  • sh/js/coverage.py
  • sh/rs/coverage.sh
  • test/wasm/baseline/driver.ts
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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.

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. Another round soon, please!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

@kixelated
kixelated added this pull request to the merge queue Oct 22, 2026
Merged via the queue into main with commit c796933 Oct 22, 2026
12 checks passed
@kixelated
kixelated deleted the quest/m1/audit-ci branch October 22, 2026 23:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant