From 7b47b89465402a7a4414382d3a90043facad1cdf Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Wed, 21 Oct 2026 20:51:11 -0700 Subject: [PATCH 1/3] quest: claim quest/m1/quic-silent-peer-drain Co-Authored-By: Claude Opus 5.5 From f4c06d765000a35bc12090cd18d0a4be2a9c6ed2 Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Wed, 21 Oct 2026 20:56:10 -0700 Subject: [PATCH 2/3] quest: close the silent-peer drain quest, landed in #6201 moq-dev/moq#6201 already reports an unanswered handshake as no longer active at its local close, so Client::close and a moqsink stop return at once. Correct moq-uring's handshake timeout doc, which named the idle timeout where moq-quic's fixed handshake idle timeout applies. Co-Authored-By: Claude Opus 5.5 --- quest/m1/README.md | 1 - quest/m1/quic-silent-peer-drain.md | 33 ------------------------------ rs/moq-uring/src/quic/handshake.rs | 7 +++++-- 3 files changed, 5 insertions(+), 36 deletions(-) delete mode 100644 quest/m1/quic-silent-peer-drain.md diff --git a/quest/m1/README.md b/quest/m1/README.md index b3b330fc544..475da4fce57 100644 --- a/quest/m1/README.md +++ b/quest/m1/README.md @@ -111,7 +111,6 @@ lite-07 stream type here. - [Received timestamps are media time](/quest/m1/lite-time-provenance.md) - a received timed track's timestamps are media time, so the rewind check can run across hops - [WebGPU on Safari](/quest/m1/webgpu-safari.md) - the WebGPU renderer is verified on Safari 26 for macOS and iOS - [Close tail drain](/quest/m1/close-tail-drain.md) - after a peer close, subscribers read the group streams their transport already delivered instead of aborting them, so a finished tail arrives on every version -- [Silent-peer close](/quest/m1/quic-silent-peer-drain.md) - moq-quic closes a connection that never heard its peer without the 3 s drain - [Binding linger](/quest/m1/linger-ffi.md) - `moq-ffi` and every binding finish a broadcast cleanly, so subscribers drain it until its media is stale - [JS linger](/quest/m1/linger-js.md) - `js/net` finishes a broadcast cleanly like Rust, and `@moq/publish` uses it when a source stops - [Gateway linger](/quest/m1/linger-gateways.md) - HLS, RTMP, SRT and RTC imports linger at a clean end, cut short by a newer epoch at the same path diff --git a/quest/m1/quic-silent-peer-drain.md b/quest/m1/quic-silent-peer-drain.md deleted file mode 100644 index 9aae84717ae..00000000000 --- a/quest/m1/quic-silent-peer-drain.md +++ /dev/null @@ -1,33 +0,0 @@ -# [S] A connection that never heard its peer closes at once - -## Goal - -moq-quic skips the close drain for a connection that never received a packet -from its peer, so closing a dial to an unreachable host returns at once -instead of waiting three PTOs (3 s at the default 333 ms initial RTT). -Stopping a `moqsink` whose relay is down stops promptly. - -## Plan - -Seen as moqsink's 3 s stop against an unreachable relay in -[moq-dev/moq#6152](https://github.com/moq-dev/moq/pull/6152), and as about ten -moq-gst tests paying 3 s each in -[moq-dev/moq#6174](https://github.com/moq-dev/moq/pull/6174). With no packet -from the peer, nothing on the other side needs the drain's CONNECTION_CLOSE -retransmits; a peer heard even once keeps the normal drain. - -Open decision, public API, settled in this quest's PR with options and a -recommendation: should the configured idle timeout also bound the handshake? -Today moq-quic's handshake idle timeout is a fixed 10 s that neither -`--quic-idle-timeout` nor moq-uring's `Transport` sets, so three moq-uring -endpoint tests wait it out (#6174). - -Regression tests: close a client whose dial never got an answer and assert -it returns without the drain, on a mocked clock. - -Wire: none. - -## Related - -- [Socket close](/quest/m2/quic-socket-close.md) - releasing the endpoint's socket on close -- [Keep-alive from the idle timeout](/quest/m2/quic-keep-alive.md) - the other timer derived from the idle timeout diff --git a/rs/moq-uring/src/quic/handshake.rs b/rs/moq-uring/src/quic/handshake.rs index 97d9646340d..1047f5f0e0b 100644 --- a/rs/moq-uring/src/quic/handshake.rs +++ b/rs/moq-uring/src/quic/handshake.rs @@ -14,8 +14,11 @@ use moq_clock::Instant; #[non_exhaustive] pub struct Config { /// How long an incoming connection has from its first packet to finish its - /// whole handshake, or `None` (the default) for no bound but the idle - /// timeout. + /// whole handshake, or `None` (the default) for no bound but moq-quic's + /// 10 s handshake idle timeout, which [`Transport::idle_timeout`] does + /// not change. + /// + /// [`Transport::idle_timeout`]: super::Transport::idle_timeout pub timeout: Option, } From c1b41e9ac7c413dd7ed534b926bcc28bc23fa4a4 Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Wed, 21 Oct 2026 21:38:48 -0700 Subject: [PATCH 3/3] doc(uring): name both inactivity bounds on an unbounded handshake Without Config::timeout, moq-quic's fixed handshake idle timeout bounds the QUIC handshake, then Transport::idle_timeout bounds HTTP/3 and MoQ SETUP on top. Co-Authored-By: Claude Opus 5.5 --- rs/moq-uring/src/quic/handshake.rs | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/rs/moq-uring/src/quic/handshake.rs b/rs/moq-uring/src/quic/handshake.rs index 1047f5f0e0b..6e0d0f61ec5 100644 --- a/rs/moq-uring/src/quic/handshake.rs +++ b/rs/moq-uring/src/quic/handshake.rs @@ -14,9 +14,11 @@ use moq_clock::Instant; #[non_exhaustive] pub struct Config { /// How long an incoming connection has from its first packet to finish its - /// whole handshake, or `None` (the default) for no bound but moq-quic's - /// 10 s handshake idle timeout, which [`Transport::idle_timeout`] does - /// not change. + /// whole handshake, or `None` (the default) for no deadline. Without one, + /// only inactivity ends a stalled handshake: moq-quic's fixed 10 s + /// handshake idle timeout until QUIC is established, which + /// [`Transport::idle_timeout`] does not change, then + /// [`Transport::idle_timeout`] through the layers on top. /// /// [`Transport::idle_timeout`]: super::Transport::idle_timeout pub timeout: Option,