Skip to content

quest(quic): record the BBR release steps and media check - #4193

Merged
kixelated merged 4 commits into
quest/m1/quic/READMEfrom
quest/m1/quic/bbr-release
Sep 25, 2026
Merged

kixelated merged 4 commits into
quest/m1/quic/READMEfrom
quest/m1/quic/bbr-release

Conversation

@kixelated

@kixelated kixelated commented Sep 25, 2026 •

Copy link
Copy Markdown
Collaborator

Updates the BBR release quest with the exact steps the maintainer runs to publish and pin the fork, and the media-shaped check the quest requires. The quest stays open until the release ships and the pin lands. Nothing here publishes.

Changes

The pin change isn't committed, because "1.3.1" can't resolve until the crates exist, so --locked would fail. It raises the floor for the fork crates MoQ names directly, so downstream consumers of the next moq-tokio and moq-uring releases get the fixed BBR:

-moq-noq-proto = { version = "1.3", default-features = false }
-moq-noq-udp = "1.3"
+moq-noq-proto = { version = "1.3.1", default-features = false }
+moq-noq-udp = "1.3.1"
-web-transport-moq = { version = "1.3", default-features = false }
+web-transport-moq = { version = "1.3.1", default-features = false }

Validation against the local fork build

  • One copy of each fork crate across --all-features (cargo tree -d).
  • clippy -D warnings, default features: moq-tokio, moq-uring, moq-relay, moq-native, moq-bench, moq-cli. With --all-features: moq-tokio, moq-uring, moq-relay. --no-default-features check, and just rs tokio-features. All pass.
  • nextest --all-features: moq-tokio and moq-relay 658/658. moq-uring: every test passed in some run, but io_uring worker setup hit ENOMEM on this shared host, on different tests each run, before any QUIC code ran.
  • Media check, QUIC only, two 60s runs per cell, one 4.8 Mbps track to two subscribers through the relay, about 71 MB offered:
    • 40ms RTT, 2% loss, 40 Mbps: 1.3.0 received 23-26 MB, p50 241-340ms. patched received 69 MB, p50 54-55ms, p99 131-141ms.
    • 40ms RTT, 1% loss, 16 Mbps: 1.3.0 received 53-55 MB. patched received 57-61 MB, p99 293-376ms.
  • The WebSocket fallback won the connect race in several first-pass runs under loss, so the check turns it off and asserts from the relay log that all three sessions are QUIC.

Public API and wire impact

None here. After the pin: the fork's Controller gains additive, space-aware callbacks. MoQ doesn't implement or call Controller, so it doesn't change. No wire change.

(Written by Claude Opus 5.5)

🤖 Generated with Claude Code

kixelated and others added 2 commits September 25, 2026 12:33
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
moq-dev/noq#10 prepares 1.4.0 of the fork. The quest now lists the parent,
upstream status, the ordered steps the maintainer runs to tag, publish, and
pin, and the netem media check with its 1.3.0 versus 1.4.0 results from a
local [patch] build.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Sep 25, 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 2026-09-25T22:37:00.403800Z 064851e 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.

@kixelated

Copy link
Copy Markdown
Collaborator Author

MERGE

Positive improvement: yes. This turns a vague “publish and pin” quest into a concrete maintainer runbook: ordered release steps, trusted-publisher caveat for web-transport-moq, exact pin/cargo update commands, the required checks, and a reproducible netem media recipe with measured 1.3.0 vs 1.4.0 outcomes. Leaving the pin uncommitted until 1.4.0 exists on crates.io is the right call (--locked would fail otherwise).

Worth the complexity: yes, and the cost is low. One markdown file, no code or lockfile churn. Recording local [patch.crates-io] numbers with the expectation that published crates match is enough to gate the release without blocking the writeup on a publish that has not happened yet.

Different approach: putting the release checklist only in moq-dev/noq#10 would lose the MoQ-side pin and media gate that this quest owns. Waiting to document until crates are live would just delay the same content. No better split stands out.

Recommendation: merge as-is. After moq-dev/noq#10 lands and v1.4.0 publishes, follow steps 4–6 in a separate pin PR and delete this quest there.

This is an automated review, not the maintainer's decision
(Written by Grok)

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 877bfe28b7

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".

Comment thread quest/m1/quic/bbr-release.md Outdated
moq-bench --connect http://localhost:4443 --connect-websocket-enabled=false \
--fanout media --connections 3 --startup 1s --duration 60s --report 5s \
--fps 30 --frame-size 20000 --group-size 29 --output stats.jsonl
kill %1'

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Preserve the benchmark failure through cleanup

If moq-bench exits nonzero, for example because the required media delivery fails, this shell still runs kill %1; a successful kill then becomes bash -c's exit status, so unshare reports success and the release check can pass despite a failed benchmark. Capture and return the benchmark status after cleanup, or use a cleanup trap that preserves it.

AGENTS.md reference: AGENTS.md:L17-L17

Useful? React with 👍 / 👎.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b304466ab6

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".

Comment on lines +69 to +71
moq-bench --connect http://localhost:4443 --connect-websocket-enabled=false \
--fanout media --connections 3 --startup 1s --duration 60s --report 5s \
--fps 30 --frame-size 20000 --group-size 29 --output stats.jsonl

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Gate the release on corrected media performance

When a BBR regression still delivers some media, as the documented 1.3.0 results do, this command exits successfully because Stats::ensure_delivery only requires groups_recv > 0; stats.jsonl is written but never checked against the 1.3.1 throughput or latency expectations. Consequently the required release gate can pass with the known-bad behavior even after the benchmark exit status is preserved. Add an explicit assertion on the recorded delivery and latency results.

AGENTS.md reference: AGENTS.md:L18-L18

Useful? React with 👍 / 👎.

Comment thread quest/m1/quic/bbr-release.md Outdated
unshare -rn bash -c '
ip link set lo up
tc qdisc add dev lo root netem delay 20ms loss 2% rate 40mbit
moq-relay relay.toml & # bind 127.0.0.1:4443, tls.generate, public auth, no iroh

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Provide the relay configuration used by the check

In a fresh checkout run from the repository root, moq-relay relay.toml cannot start because that file is neither created by the snippet nor present at this path. A repo-wide search finds files with that name only under packaging/moq-relay and test/wasm, and neither is the configuration described by the inline comment. Point to a checked-in suitable config or create the exact config in the snippet so the release check is reproducible.

Useful? React with 👍 / 👎.

Point the relay at test/interop/interop.toml, keep moq-bench's exit status
across the cleanup kill, and reject the 1.3.0 delivery shape on the 40 Mbps
path.

Co-authored-by: Grok 4.7 <noreply@x.ai>
@kixelated

Copy link
Copy Markdown
Collaborator Author

Landing the BBR release runbook on the quic quest line. It records the 1.3.1 publish steps and the netem media results.

The media-check snippet now fails closed:

  • the relay config is test/interop/interop.toml (127.0.0.1:4443, generated cert, public auth, no iroh)
  • the shell keeps moq-bench's exit status, so kill cannot hide a failed run
  • the 40 Mbps path rejects a result that still looks like 1.3.0 (under 60 MB, p50 over 150 ms, or p99 over 400 ms)

The 16 Mbps row stays a table comparison. 1.3.0 already delivers most of the bytes there and the latency ranges overlap, so those cutoffs would not separate the two builds.

(Written by Grok 4.7)

@kixelated
kixelated merged commit 5fa6bf9 into quest/m1/quic/README Sep 25, 2026
3 checks passed
@kixelated
kixelated deleted the quest/m1/quic/bbr-release branch September 25, 2026 22:34

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 064851e8fa

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".

unshare -rn bash -c '
ip link set lo up
tc qdisc add dev lo root netem delay 20ms loss 2% rate 40mbit
moq-relay test/interop/interop.toml &

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Run the checkout's release binaries explicitly

In a fresh checkout using the repository-required Nix development shell, neither moq-relay nor moq-bench is on PATH, and the preceding clippy and nextest commands do not create release executables. The documented release gate therefore stops here, or can accidentally test an unrelated globally installed version instead of the newly locked crates. Add an explicit release build and invoke target/release/moq-relay and target/release/moq-bench. (Written by GPT-5.6 Sol)

AGENTS.md reference: AGENTS.md:L82-L83

Useful? React with 👍 / 👎.

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