Add support for GOAWAY draining events to support WebTransport

[jesup]

No description provided.

[Copilot]

Pull request overview

Adds client-side WebTransport “draining” notifications in response to HTTP/3 GOAWAY, aligning behavior with the WebTransport-over-HTTP/3 spec’s graceful shutdown expectations.

Changes:

• Emit WebTransportEvent::Draining { session_id } for all WebTransport sessions when the client receives GOAWAY.
• Add helper APIs to identify WebTransport sessions and to inject GOAWAY frames in tests.
• Add WebTransport tests covering GOAWAY behavior for both “active” and “rejected” sessions.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
neqo-http3/src/server.rs	Adds a test-only helper to queue GOAWAY on all server connections.
neqo-http3/src/connection_server.rs	Adds a test-only helper to queue a GOAWAY control frame on a server handler.
neqo-http3/src/connection_client.rs	Emits WebTransportEvent::Draining when handling GOAWAY, before resetting rejected streams.
neqo-http3/src/connection.rs	Adds webtransport_session_ids() helper to enumerate WebTransport session stream IDs.
neqo-http3/src/client_events.rs	Introduces the new public WebTransportEvent::Draining variant.
neqo-http3/src/features/extended_connect/session.rs	Exposes connect_type() on extended-connect sessions for filtering session kinds.
neqo-http3/src/features/extended_connect/webtransport_session.rs	Adds (currently unused) draining state and related accessors/TODOs.
neqo-http3/src/features/extended_connect/tests/webtransport/sessions.rs	Adds tests validating Draining + (non-)closure semantics around GOAWAY.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (users/jesup/exportkey@0626e18). Learn more about missing BASE report.

Additional details and impacted files

@@                   Coverage Diff                    @@
##             users/jesup/exportkey    #3465   +/-   ##
========================================================
  Coverage                         ?   94.38%           
========================================================
  Files                            ?      133           
  Lines                            ?    40275           
  Branches                         ?    40275           
========================================================
  Hits                             ?    38015           
  Misses                           ?     1408           
  Partials                         ?      852           

Flag	Coverage Δ
freebsd	78.55% <ø> (?)
linux	94.52% <100.00%> (?)
macos	94.46% <100.00%> (?)
windows	94.52% <100.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
neqo-common	98.61% <0.00%> (?)
neqo-crypto	86.49% <0.00%> (?)
neqo-http3	93.94% <0.00%> (?)
neqo-qpack	95.14% <0.00%> (?)
neqo-transport	95.60% <0.00%> (?)
neqo-udp	84.90% <0.00%> (?)
mtu	86.61% <0.00%> (?)

[mxinden]

I wonder whether a single draining event code path isn't a bit cleaner and easier for the consuming application. See comment below.

[jesup]

This PR is part of a stack of 13 bookmarks:

1. trunk()
2. users/jesup/exportkey
3. users/jesup/draining ← this PR
4. users/jesup/sub_protocol
5. users/jesup/sendGroup
6. users/jesup/stats
7. users/jesup/highwatermarks
8. users/jesup/sendgroup_sendorder_tracking
9. users/jesup/stats_overhead
10. users/jesup/avoid_PMTUD
11. users/jesup/anticipated_streams
12. users/jesup/waitUntilAvailable
13. users/jesup/atomicWrite
14. users/jesup/WT_DRAIN_SESSION

---

Created with jj-stack

[jesup]

ignore the downgrade bit, sorry. annoyances building; I have a temp patch for that

[codspeed-hq]

Merging this PR will not alter performance

✅ 24 untouched benchmarks
⏩ 27 skipped benchmarks¹

---

_{Comparing users/jesup/draining (1f45777) with users/jesup/exportkey (b21b342)²}

Footnotes

1. 27 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on users/jesup/exportkey (0626e18) during the generation of this report, so 7e1e1ba was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[github-actions]

Benchmark results

Significant performance differences relative to 0626e18.

streams/walltime/1000-streams/each-1000-bytes: 💔 Performance has regressed by +1.3111%.

       time:   [45.146 ms 45.191 ms 45.237 ms]
       change: [+1.0108% +1.3111% +1.5450] (p = 0.00 < 0.05)
       Performance has regressed.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high mild

All results

transfer/1-conn/1-100mb-resp (aka. Download): No change in performance detected.

       time:   [211.82 ms 212.26 ms 212.79 ms]
       thrpt:  [469.95 MiB/s 471.13 MiB/s 472.10 MiB/s]
change:
       time:   [-0.1417% +0.1860% +0.5078] (p = 0.27 > 0.05)
       thrpt:  [-0.5052% -0.1856% +0.1419]
       No change in performance detected.
Found 2 outliers among 100 measurements (2.00%)
2 (2.00%) high severe

transfer/1-conn/10_000-parallel-1b-resp (aka. RPS): Change within noise threshold.

       time:   [286.22 ms 287.79 ms 289.35 ms]
       thrpt:  [34.560 Kelem/s 34.747 Kelem/s 34.939 Kelem/s]
change:
       time:   [-1.9362% -1.0431% -0.1755] (p = 0.02 < 0.05)
       thrpt:  [+0.1758% +1.0541% +1.9744]
       Change within noise threshold.

transfer/1-conn/1-1b-resp (aka. HPS): No change in performance detected.

       time:   [38.681 ms 38.828 ms 38.997 ms]
       thrpt:  [25.643   B/s 25.754   B/s 25.852   B/s]
change:
       time:   [-0.6048% +0.0244% +0.5989] (p = 0.94 > 0.05)
       thrpt:  [-0.5954% -0.0244% +0.6085]
       No change in performance detected.
Found 6 outliers among 100 measurements (6.00%)
6 (6.00%) high severe

transfer/1-conn/1-100mb-req (aka. Upload): Change within noise threshold.

       time:   [213.28 ms 213.64 ms 213.99 ms]
       thrpt:  [467.32 MiB/s 468.08 MiB/s 468.86 MiB/s]
change:
       time:   [+0.7063% +0.9453% +1.1777] (p = 0.00 < 0.05)
       thrpt:  [-1.1640% -0.9364% -0.7014]
       Change within noise threshold.

streams/walltime/1-streams/each-1000-bytes: No change in performance detected.

       time:   [588.05 µs 590.65 µs 593.65 µs]
       change: [-0.6882% -0.0395% +0.6462] (p = 0.91 > 0.05)
       No change in performance detected.
Found 11 outliers among 100 measurements (11.00%)
11 (11.00%) high severe

streams/walltime/1000-streams/each-1-bytes: Change within noise threshold.

       time:   [12.357 ms 12.384 ms 12.416 ms]
       change: [+0.5378% +0.8241% +1.1448] (p = 0.00 < 0.05)
       Change within noise threshold.
Found 2 outliers among 100 measurements (2.00%)
1 (1.00%) high mild
1 (1.00%) high severe

streams/walltime/1000-streams/each-1000-bytes: 💔 Performance has regressed by +1.3111%.

       time:   [45.146 ms 45.191 ms 45.237 ms]
       change: [+1.0108% +1.3111% +1.5450] (p = 0.00 < 0.05)
       Performance has regressed.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high mild

transfer/walltime/pacing-false/varying-seeds: Change within noise threshold.

       time:   [23.099 ms 23.129 ms 23.176 ms]
       change: [-1.7939% -1.6112% -1.3952] (p = 0.00 < 0.05)
       Change within noise threshold.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high severe

transfer/walltime/pacing-true/varying-seeds: Change within noise threshold.

       time:   [23.492 ms 23.512 ms 23.532 ms]
       change: [+0.0106% +0.1300% +0.2503] (p = 0.03 < 0.05)
       Change within noise threshold.
Found 3 outliers among 100 measurements (3.00%)
3 (3.00%) high mild

transfer/walltime/pacing-false/same-seed: Change within noise threshold.

       time:   [23.562 ms 23.583 ms 23.606 ms]
       change: [-0.9367% -0.8229% -0.7073] (p = 0.00 < 0.05)
       Change within noise threshold.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high severe

transfer/walltime/pacing-true/same-seed: No change in performance detected.

       time:   [23.859 ms 23.883 ms 23.911 ms]
       change: [-0.1826% -0.0170% +0.1460] (p = 0.84 > 0.05)
       No change in performance detected.
Found 6 outliers among 100 measurements (6.00%)
1 (1.00%) low mild
4 (4.00%) high mild
1 (1.00%) high severe

Download data for profiler.firefox.com or download performance comparison data.

[github-actions]

Client/server transfer results

Performance differences relative to 0626e18.

Transfer of 33554432 bytes over loopback, min. 100 runs. All unit-less numbers are in milliseconds.

Client vs. server (params)	Mean ± σ	Min	Max	MiB/s ± σ	Δ baseline	Δ baseline
neqo-msquic-cubic	150.7 ± 4.9	143.7	173.1	212.3 ± 6.5	💚 -3.4	-2.2%
neqo-neqo-cubic	90.7 ± 2.5	86.7	97.5	352.7 ± 12.8	💔 0.8	0.9%
neqo-neqo-newreno	90.6 ± 2.6	86.5	98.8	353.1 ± 12.3	💔 0.7	0.8%

Table above only shows statistically significant changes. See all results below.

All results

Transfer of 33554432 bytes over loopback, min. 100 runs. All unit-less numbers are in milliseconds.

Client vs. server (params)	Mean ± σ	Min	Max	MiB/s ± σ	Δ baseline	Δ baseline
google-google-nopacing	455.0 ± 1.8	450.5	462.3	70.3 ± 17.8
google-neqo-cubic	266.8 ± 2.6	261.8	279.1	120.0 ± 12.3	-0.0	-0.0%
msquic-msquic-nopacing	132.7 ± 38.4	111.1	356.7	241.2 ± 0.8
msquic-neqo-cubic	163.9 ± 44.8	122.9	356.5	195.3 ± 0.7	5.5	3.5%
neqo-google-cubic	768.0 ± 2.3	762.8	775.0	41.7 ± 13.9	0.2	0.0%
neqo-msquic-cubic	150.7 ± 4.9	143.7	173.1	212.3 ± 6.5	💚 -3.4	-2.2%
neqo-neqo-cubic	90.7 ± 2.5	86.7	97.5	352.7 ± 12.8	💔 0.8	0.9%
neqo-neqo-cubic-nopacing	89.4 ± 2.6	85.7	96.4	358.0 ± 12.3	0.1	0.1%
neqo-neqo-newreno	90.6 ± 2.6	86.5	98.8	353.1 ± 12.3	💔 0.7	0.8%
neqo-neqo-newreno-nopacing	90.5 ± 2.9	84.6	99.2	353.5 ± 11.0	0.5	0.6%
neqo-quiche-cubic	190.0 ± 3.6	185.3	213.8	168.4 ± 8.9	0.5	0.3%
neqo-s2n-cubic	215.8 ± 2.0	209.9	221.1	148.3 ± 16.0	0.2	0.1%
quiche-neqo-cubic	170.7 ± 2.9	165.1	179.4	187.4 ± 11.0	-0.7	-0.4%
quiche-quiche-nopacing	138.9 ± 2.8	133.5	151.3	230.3 ± 11.4
s2n-neqo-cubic	213.5 ± 3.1	207.3	224.1	149.9 ± 10.3	-0.6	-0.3%
s2n-s2n-nopacing	292.7 ± 25.5	276.0	388.1	109.3 ± 1.3

Download data for profiler.firefox.com or download performance comparison data.

[github-actions]

Failed Interop Tests

None ❓

All results

Succeeded Interop Tests

QUIC Interop Runner, client vs. server

neqo-pr as client

None ❓

neqo-pr as server

None ❓

Unsupported Interop Tests

QUIC Interop Runner, client vs. server

neqo-pr as client

None ❓

neqo-pr as server

None ❓