[kixelated]

Problem

The nightly competing_cubic probe test treated stream completion as a reset. After a stream was acknowledged or freed, its expiry helper recorded the group's end time as a fake reset time. It then rejected older probe packets that were still in flight.

Approach

Record a reset time only when the helper actually resets the stream. The assertion against resending after a real reset stays, and every acceptance budget is unchanged.

This PR does not finish the quest. With the bookkeeping fixed, competing_cubic now gets far enough to fail its deadline budget. quest/m0/nightly/competing-cubic.md is rewritten to cover only that remaining lateness work, and the quest stays open until it is fixed.

Impact

• Public API and wire: none. This changes test bookkeeping only; production probing is unchanged.

Alternatives

Changing which packets count as probes, or adding another group lifecycle state, would add complexity and still leave the false timestamp in place. Removing the fake reset time fixes the assertion failure that was confirmed.

Choices

• ✅ Accept this isolated test fix now and keep the quest open for the deadline-budget lateness beside Cubic (the user's decision). The alternative was to hold this PR until both failures were fixed together.
• Merged main after moq-dev/moq#6850 landed (no rebase). Its quest edits had been squash-merged, so conflicts resolved to main's version. The only exception is this PR's own quest.
• The nightly epic's link to this quest now describes the remaining deadline-budget work.

Validation

• Reproduced the original release failure. Temporary diagnostics showed the rejected probe was sent before the stream completed. The diagnostics were removed.
• After the merge, ran the release probe suite with debug assertions and overflow checks and --no-fail-fast, as just rs quic-sim does. 12 of 13 passed. competing_cubic passes the reset assertion and fails only its deadline budget (probe.rs:596): over seeds 1 through 8, probing averages 413.6 late frames against 385.1 for the baseline. The budget is unchanged, and that failure is what the quest now covers.
• After the merge, just check passed (Rust: 5976 tests, all passed) and quest check --base origin/main passed.

Follow-ups

• quest/m0/nightly/competing-cubic.md stays open: find why probing beside Cubic adds delay and fix it in the transport.

(Written by Claude Opus 5.5)

[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 2 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: cf9f4727-d706-472d-aab2-e40723e0c886

Commits

Reviewing files that changed from the base of the PR and between 802aa29 and e463f60.

Files selected for processing: 3

• quest/m0/nightly/README.md
• quest/m0/nightly/competing-cubic.md
• rs/moq-quic/src/tests/probe.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 2026-10-26T18:44:19.023741Z	e463f60	Draft marked ready

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.

[kixelated]

Automated review by review (OpenAI)

Reviewed head e463f60a328389e590950253997b2e4217f6060c against base 802aa29d7f686e09aaf21dade7dd6e048b704258.

No new actionable findings. In expire, leaving reset unset for acknowledged/freed streams correctly avoids comparing still-in-flight probes against an invented reset time. Actual resets still record now and retain the no-resend assertion. This one-line bookkeeping correction is the simpler direction; production probe behavior and acceptance budgets remain unchanged.

The documented Cubic deadline-budget failure remains open in the quest; this is not a claim that the nightly suite is green. No prior review findings to reconcile; the independent Codex summary also reports no findings.

Verification: GitHub-only static inspection of all three changed files, the surrounding test lifecycle, offset/acked/reset handling, and in-flight probe enumeration. I did not run tests or independently reproduce the reported lateness.

[kixelated]

Landing e463f60a328389e590950253997b2e4217f6060c via the merge queue (squash).

• Changes: the probe test's expiry helper records a reset time only on a real reset, so acknowledged or freed streams no longer reject still-in-flight probes. Production probing and acceptance budgets are unchanged.
• Decision: land this isolated test fix now (the user's pick); quest/m0/nightly/competing-cubic.md stays open for the deadline-budget lateness beside Cubic, so nightly just rs quic-sim keeps failing on competing_cubic until that lands.
• Reviews on this head: Codex (no findings) and the OpenAI review (no actionable findings). CI passed; quest check --base origin/main passed on the head.

(Written by Claude Opus 5.5)