[kixelated]

Problem

The exact flapping-peer seed 783778834340462316 still failed on main e900b9f62a3c63ba577811974f45119f3251194e, after moq-dev/moq#5943. Recovery succeeded, but impairment verification counted lost and overflowing datagrams as chances for reorder to act. The failing path had 499 datagrams but only 436 could apply delay or reorder.

Approach

Count loss exposure separately from admission to delay/reorder, preserving each decision's phase. Keep the existing probability cutoff, seeded draws, impairment profile, and delivery assertions. Add deterministic recorded-count and actual admission-path regressions, then remove the completed quest and its reference.

Decisions

• Merge origin/main to bring the branch up to date: ✅ merge (no conflicts) / rebase (never, per the repo rules).
• Draft state: ✅ mark ready, since nothing is unfinished and the earlier draft hold was a no-merge instruction, which ready-for-review does not change / keep draft.

Validation

• Exact command in the Nix shell: MOQ_SHAPER_SEED=783778834340462316 just test drill -E 'test(=bursts_cross_a_flapping_peer::impaired)' --success-output immediate.
• Baseline: exit 100, original reorder-verification failure. The deterministic regression also failed before the fix.
• Same seed after the source fix: exit 0, 1 passed in 16.299s. All 120 groups recovered, with flap and bottleneck overflow asserted. Scheduling still changes traffic, so this is a single replay result.
• Frozen head d1907360df980f7620caf36648b0a87f7f5f6bf1: just check e900b9f62a3c63ba577811974f45119f3251194e passed.
• After a first origin/main merge (head 7184eea474589e5d4fffad17482d3390e745f573): the seed replay passed again (16.5s), and just check passed every gate (644 + 216 tests, lint, quest, drill mutations, flake) except bench/relay.test.py's test_a_foreign_listener_never_proves_relay_readiness. That fixture is untouched here and fails the same on main on this host: its hold.py picks a random port that collided with a busy listener (EADDRINUSE), so the harness blocks forever reading the never-written .ready FIFO.
• After a second origin/main merge (head 07382b2563, quest README conflict resolved by keeping main's new macOS QUIC child and dropping this completed quest's entry): just check passed in full, including bench/relay.test.py.

Impact

• Public API: none.
• Wire: none.

Alternatives

Changing the seed, profile, cutoff, timeouts, or delivery assertions would hide the accounting bug.

Follow-ups

• bench/relay.test.py foreign-listener fixture: a port collision in hold.py hangs the test instead of retrying or failing, which breaks just check on loaded hosts.

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

[kixelated]

Quest outcome: the exact seed reproduced the original verifier failure on e900b9f62a3c63ba577811974f45119f3251194e (exit 100), and a deterministic regression confirmed dropped datagrams were counted as reorder opportunities. The corrected per-phase accounting passes that seed (exit 0) and the complete required changed-package gate on d1907360df980f7620caf36648b0a87f7f5f6bf1 (exit 0). Evidence is preserved in the isolated worktree's .scratch/.

No public API or wire changes, open implementation decisions, or remaining local checks. All owned commands have ended. The PR remains draft under the user's explicit no-merge instruction.

(written by GPT-6.1-Sol)

[coderabbitai]

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 4 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

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 2191f7e8-25b3-4606-8713-fac4230b25df

📥 Commits

Reviewing files that changed from the base of the PR and between 697441d and 07382b2.

📒 Files selected for processing: 3

• quest/m1/test-flakes-2/README.md
• quest/m1/test-flakes-2/flapping-peer-reorder.md
• rs/moq-shaper/src/lib.rs

• 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

• X
• Mastodon
• Reddit
• LinkedIn

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

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-20T08:59:09.683916Z	07382b2	New commits

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.