feat: Retain final path stats if a `Path` is alive, add `WeakPathHandle` by Frando · Pull Request #386 · n0-computer/noq

Frando · 2026-01-30T13:18:04Z

Description

Once a path is abandoned, its state is immediately dropped from quinn_proto::Connection. The final path stats are emitted as an event, which is handled in quinn::Connection and forwarded over the path events broadcast channel.

This construction makes sure that no state is kept around for abandoned paths to prevent memory leaks. However, it gives consumers a hard time, as we experience in iroh with the difficulties in accessing the final path stats.

By sending the final path stats over a broadcast channel while they are already removed from the underlying connection, consumers of the API have this weird in-between state where the final stats for an abandoned path are not accessible for the period of time between t1: remove stats from connection and send into broadcats channel and t2: consumer of path event broadcast channel receives the event. Between t1 and t2, the final stats are not accessible at all.

This PR changes quinn such that within a quinn::Connection we keep a reference counter for Path structs currently alive. As long as a Path struct is alive, once a path is abandoned we store the path stats within the quinn::Connection until all references to this path are dropped.

This means that Path::stats now can unconditionally return the stats for that path, even if the path was abandoned in the meantime.

Great! However, still not enough for iroh. Keeping a quinn::Path struct around prevents the quinn::Connection from being closed-on-drop. In iroh, we want a "passive" path watcher which does not keep the connection alive, but still ensures that path stats for the watched paths are not dropped.

Therefore, this PR additionally adds a WeakPathHandle. It contains a WeakConnectionHandle and a PathId, so it does not prevent a connection from being closed-on-drop. However, it does take part in the reference counting of path refs, and thus does prevent the stats for abandoned paths to be dropped as long as there is a WeakPathHandle alive.

Breaking Changes

Notes & open questions

If there are ideas on how to handle this simpler, I'm all ears. I spent quite some time thinking this through, and this is the best I could come up so far.

github-actions · 2026-01-30T13:20:18Z

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/quinn/pr/386/docs/iroh_quinn/

Last updated: 2026-02-06T19:32:15Z

codecov-commenter · 2026-01-30T13:22:20Z

Codecov Report

❌ Patch coverage is 65.54622% with 41 lines in your changes missing coverage. Please review.
✅ Project coverage is 77.34%. Comparing base (c332d42) to head (93d27db).
⚠️ Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
quinn/src/path.rs	58.75%	33 Missing ⚠️
quinn/src/connection.rs	78.37%	8 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #386      +/-   ##
==========================================
+ Coverage   76.97%   77.34%   +0.36%     
==========================================
  Files          81       81              
  Lines       23030    23154     +124     
==========================================
+ Hits        17727    17908     +181     
+ Misses       5303     5246      -57

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

github-actions · 2026-01-30T13:57:38Z

Performance Comparison Report

0f0b8408a84bab7794eb518a6e4fa0fce1ce7503 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5346.6 Mbps	N/A	N/A	91.9% / 97.6%
medium-concurrent	5336.1 Mbps	N/A	N/A	90.8% / 96.9%
medium-single	3792.1 Mbps	N/A	N/A	88.6% / 97.3%
small-concurrent	3706.4 Mbps	N/A	N/A	90.2% / 99.1%
small-single	3304.4 Mbps	N/A	N/A	90.6% / 97.9%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2833.5 Mbps	3714.3 Mbps	-23.7%
lan	768.4 Mbps	796.4 Mbps	-3.5%
lossy	55.9 Mbps	N/A	N/A
wan	83.8 Mbps	N/A	N/A

Summary

iroh-quinn is 20.1% slower on average

---

2f9dafca52a3f874e17fbccfc72516e7eda4b1d0 - artifacts

No results available

---

062a68a079d24d0f4105fcf46e9c29311da15bac - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5378.1 Mbps	7862.4 Mbps	-31.6%	92.8% / 97.9%
medium-concurrent	5312.2 Mbps	7779.4 Mbps	-31.7%	91.8% / 97.0%
medium-single	3916.4 Mbps	4487.5 Mbps	-12.7%	90.3% / 98.1%
small-concurrent	3734.3 Mbps	5274.3 Mbps	-29.2%	93.3% / 103.0%
small-single	3338.3 Mbps	4734.8 Mbps	-29.5%	92.9% / 102.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2987.5 Mbps	3685.6 Mbps	-18.9%
lan	782.4 Mbps	796.4 Mbps	-1.8%
lossy	69.8 Mbps	69.9 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 26.4% slower on average

---

1fc892fd06bf57fe328e64d1cc276d75baa42da7 - artifacts

No results available

---

e52d84a2fb6c3b521e42034852fea0d27c74d2fb - artifacts

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2966.3 Mbps	3603.1 Mbps	-17.7%

Summary

iroh-quinn is 17.7% slower on average

---

e4c3a5db6b2eb18ce04d8ea60b918a9f8b076af6 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5455.2 Mbps	7752.3 Mbps	-29.6%	93.2% / 97.9%
medium-concurrent	5349.2 Mbps	7026.4 Mbps	-23.9%	91.0% / 96.9%
medium-single	3904.3 Mbps	4020.8 Mbps	-2.9%	99.9% / 187.0%
small-concurrent	3729.8 Mbps	5081.4 Mbps	-26.6%	91.0% / 99.1%
small-single	3293.0 Mbps	4389.0 Mbps	-25.0%	85.3% / 95.9%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2862.0 Mbps	3929.1 Mbps	-27.2%
lan	768.5 Mbps	804.8 Mbps	-4.5%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 23.0% slower on average

---

3475d3be1b3cf0b196c9c0becfcaaa5ccd6996ac - artifacts

No results available

---

37f697eb5ba52913e248d571c2d8f4a80f38a012 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5356.0 Mbps	8004.1 Mbps	-33.1%	94.3% / 111.0%
medium-concurrent	5395.0 Mbps	7763.4 Mbps	-30.5%	92.7% / 97.5%
medium-single	3791.7 Mbps	4189.4 Mbps	-9.5%	99.8% / 168.0%
small-concurrent	3629.2 Mbps	4812.6 Mbps	-24.6%	97.3% / 115.0%
small-single	3305.4 Mbps	4360.8 Mbps	-24.2%	96.7% / 115.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2835.4 Mbps	3904.5 Mbps	-27.4%
lan	768.8 Mbps	810.3 Mbps	-5.1%
lossy	55.8 Mbps	69.8 Mbps	-20.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 25.8% slower on average

---

c932bdddc24887a3623599512d65149040df1612 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5341.8 Mbps	7518.5 Mbps	-29.0%	98.0% / 178.0%
medium-concurrent	5460.6 Mbps	7287.4 Mbps	-25.1%	95.3% / 118.0%
medium-single	3860.7 Mbps	4307.7 Mbps	-10.4%	93.4% / 119.0%
small-concurrent	3877.1 Mbps	4989.0 Mbps	-22.3%	92.7% / 100.0%
small-single	3472.7 Mbps	4364.9 Mbps	-20.4%	93.5% / 122.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2792.0 Mbps	3631.4 Mbps	-23.1%
lan	768.5 Mbps	796.5 Mbps	-3.5%
lossy	69.8 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 22.2% slower on average

---

52265737605d7a214551abb43f3dfc968036fa68 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5320.7 Mbps	7939.8 Mbps	-33.0%	93.3% / 101.0%
medium-concurrent	5441.0 Mbps	7801.7 Mbps	-30.3%	90.8% / 96.7%
medium-single	3563.1 Mbps	4725.7 Mbps	-24.6%	92.0% / 104.0%
small-concurrent	3711.0 Mbps	5205.0 Mbps	-28.7%	102.7% / 194.0%
small-single	3335.0 Mbps	4751.1 Mbps	-29.8%	89.2% / 97.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2807.5 Mbps	3897.0 Mbps	-28.0%
lan	768.4 Mbps	796.5 Mbps	-3.5%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 28.8% slower on average

---

6ce9604944e949e18cf6e39cf64bd829d0e270f1 - artifacts

No results available

---

7f68a3ec97150733100894ae0d812f12d869f035 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5638.9 Mbps	8105.9 Mbps	-30.4%	95.7% / 100.0%
medium-concurrent	5636.6 Mbps	7515.7 Mbps	-25.0%	95.6% / 101.0%
medium-single	4153.8 Mbps	4363.6 Mbps	-4.8%	98.2% / 151.0%
small-concurrent	3938.1 Mbps	4954.8 Mbps	-20.5%	94.0% / 104.0%
small-single	3579.7 Mbps	4464.5 Mbps	-19.8%	93.4% / 103.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2873.2 Mbps	4009.2 Mbps	-28.3%
lan	782.4 Mbps	810.3 Mbps	-3.4%
lossy	69.8 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 22.2% slower on average

---

e2007d41fafad049c813effe08ad6c2e3e4eefa0 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5249.2 Mbps	8088.4 Mbps	-35.1%	95.8% / 100.0%
medium-concurrent	5277.2 Mbps	7802.1 Mbps	-32.4%	95.0% / 101.0%
medium-single	4059.7 Mbps	4749.8 Mbps	-14.5%	99.2% / 153.0%
small-concurrent	3738.5 Mbps	4999.8 Mbps	-25.2%	91.4% / 99.4%
small-single	3453.9 Mbps	4836.0 Mbps	-28.6%	94.3% / 104.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2841.5 Mbps	4090.2 Mbps	-30.5%
lan	770.2 Mbps	810.3 Mbps	-5.0%
lossy	69.9 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 28.1% slower on average

---

93d27db89f3b069ba32eda9e55e1575a250ac595 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5923.2 Mbps	7945.0 Mbps	-25.4%	97.2% / 98.8%
medium-concurrent	5401.1 Mbps	7623.9 Mbps	-29.2%	95.5% / 97.4%
medium-single	4172.4 Mbps	4749.0 Mbps	-12.1%	96.5% / 98.5%
small-concurrent	3669.5 Mbps	5359.0 Mbps	-31.5%	96.5% / 98.8%
small-single	3561.7 Mbps	4748.8 Mbps	-25.0%	96.1% / 98.3%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2873.4 Mbps	4101.9 Mbps	-30.0%
lan	782.3 Mbps	810.3 Mbps	-3.5%
lossy	69.9 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 25.2% slower on average

quinn/src/connection.rs

quinn/src/path.rs

flub

Generally looks fine I think. I also tried to look for some alternative for the expect of the path stats, but it is pretty difficult to do so.

flub · 2026-02-04T17:52:21Z

quinn-proto/src/connection/mod.rs

+        let path_stats = self.path_stats(path_id).unwrap_or_default();
+        self.path_stats.remove(&path_id);


I'm not sure why this needs to move, this whole function has &mut self so moving this a few lines does nothing?
(and also not sure why it's on two lines now..)

Self::path_stats sets the current RTT onto the stats, whereas the PathStats returned from remove will always have an RTT of 0.

flub · 2026-02-05T10:30:51Z

quinn/src/connection.rs

+    }
+
+    /// Returns `true` if the two [`WeakConnectionHandle`] point at the same connection.
+    pub fn ptr_eq(&self, other: &Self) -> bool {


Bikeshedding alert: not sure I like this much as a public API. Why do folks suddenly have to worry about pointers?

But also what to suggest? is_same_connection?

Also not entirely sure how much I dislike this, if more folks like this that would be fine by me. I tried to find some precedent for this kind of API in other libs but didn't find any.

Arc::ptr_eq is where I took the name from. Changing to is_same_connection sgtm.

Renamed it to is_same_conn

quinn/src/connection.rs

flub · 2026-02-05T10:36:57Z

quinn/src/connection.rs

+            .or_else(|| self.final_path_stats.get(&path_id).copied())
+    }
+
+    /// Increment the application-level reference counter for a path.


"the application" is usually the user of quinn, so "application-level" is a bit confusing to me here. Maybe "increments the [Connection]s reference counter for a path" works?

Rephrased the comment

quinn/src/tests.rs

quinn/src/connection.rs

Co-authored-by: Floris Bruynooghe <flub@n0.computer>

…watcher (#3899) ## Description Improves the path watcher for connections to reliably allow accessing the stats of abandoned paths. * Abandoned paths are not removed from the value of the watchable for a connection's paths, as long as there are active watchers (subscribers) * Only once all watchers are dropped, we drop the abandoned paths from the watchable's value on the next update * This means that if you keep a `PathWatcher` alive for the duration of a connection, it will contain *all* paths the connection ever used. * We use the changes from n0-computer/noq#386 and keep a `WeakPathHandle` for all paths in the watchable's value. This prevents the path stats to be dropped within a quinn connection even if the path is abandoned. * We do no longer store path stats anywhere in iroh directly, but only ever access them from quinn. Due to keeping the `WeakPathHandle`, we can ensure that the stats are always available as long as the connection hasn't been dropped. The combination of all this gives us a reliable way to access final path stats for all paths used in a connection, as long as you keep a reference to the connection around, which is quite straightforward to do and document. ## Breaking Changes * `Connection::paths` and `ConnectionInfo::paths` now return a `PathWatcher` (which still implements `n0_watcher::Watcher` but now also is a named struct) * `PathInfo::stats` and `PathInfo::rtt` now return `Option`. They return `None` if the underlying connection has been droped. ## Notes & open questions I spend quite some time going back-and-forth over different approaches. Very open to other ideas, but I'm a bit out of further ideas currently and this is the best I could come up with so far. ## Change checklist  - [ ] Self-review. - [ ] Documentation updates following the [style guide](https://rust-lang.github.io/rfcs/1574-more-api-documentation-conventions.html#appendix-a-full-conventions-text), if relevant. - [ ] Tests if relevant. - [ ] All breaking changes documented. - [ ] List all breaking changes in the above "Breaking Changes" section. - [ ] Open an issue or PR on any number0 repos that are affected by this breaking change. Give guidance on how the updates should be handled or do the actual updates themselves. The major ones are: - [ ] [`quic-rpc`](https://github.com/n0-computer/quic-rpc) - [ ] [`iroh-gossip`](https://github.com/n0-computer/iroh-gossip) - [ ] [`iroh-blobs`](https://github.com/n0-computer/iroh-blobs) - [ ] [`dumbpipe`](https://github.com/n0-computer/dumbpipe) - [ ] [`sendme`](https://github.com/n0-computer/sendme)

Frando mentioned this pull request Jan 30, 2026

feat(iroh)!: Retain stats for closed and abandoned paths in the path watcher n0-computer/iroh#3899

Merged

11 tasks

n0bot bot added this to iroh Jan 30, 2026

github-project-automation bot moved this to 🚑 Needs Triage in iroh Jan 30, 2026

Frando force-pushed the Frando/pathstats branch from 2f9dafc to 062a68a Compare January 30, 2026 13:58

Frando added 4 commits February 3, 2026 10:22

fix: update final path stats before emitting in event

49a034a

wip: keep final stats alive as long as Path or WeakPathHandle exists

7a01f8f

add Clone for Path

62be001

docs

e52d84a

Frando force-pushed the Frando/pathstats branch from 1fc892f to e52d84a Compare February 3, 2026 09:22

chore: clippy

e4c3a5d

Frando changed the title ~~wip: Retain final path stats if a Path is alive, add WeakPathHandle~~ feat: Retain final path stats if a Path is alive, add WeakPathHandle Feb 3, 2026

dignifiedquire moved this from 🚑 Needs Triage to 🏗 In progress in iroh Feb 3, 2026

Frando requested a review from flub February 3, 2026 11:35

Frando added 2 commits February 3, 2026 13:06

fix: path ref counter increment on clone

3475d3b

chore: fmt

37f697e

dignifiedquire reviewed Feb 3, 2026

View reviewed changes

quinn/src/connection.rs Show resolved Hide resolved

dignifiedquire reviewed Feb 3, 2026

View reviewed changes

quinn/src/connection.rs Show resolved Hide resolved

dignifiedquire reviewed Feb 3, 2026

View reviewed changes

quinn/src/path.rs Outdated Show resolved Hide resolved

dignifiedquire reviewed Feb 3, 2026

View reviewed changes

quinn/src/path.rs Show resolved Hide resolved

Frando added 3 commits February 3, 2026 14:22

add docs and test

c932bdd

chore: clippy

5226573

Merge branch 'main' into Frando/pathstats

6ce9604

Frando mentioned this pull request Feb 3, 2026

bug: PathStats seem to be missing/outdated sometimes n0-computer/iroh#3846

Closed

Frando requested a review from dignifiedquire February 5, 2026 08:40

flub requested changes Feb 5, 2026

View reviewed changes

divagant-martian assigned Frando Feb 5, 2026

Merge remote-tracking branch 'origin/main' into Frando/pathstats

d96b838

address review

7f68a3e

Frando requested a review from flub February 6, 2026 10:49

fixup test

e2007d4

flub approved these changes Feb 6, 2026

View reviewed changes

quinn/src/connection.rs Show resolved Hide resolved

Update quinn/src/connection.rs

93d27db

Co-authored-by: Floris Bruynooghe <flub@n0.computer>

Frando added this pull request to the merge queue Feb 9, 2026

Merged via the queue into main with commit b068cda Feb 9, 2026
36 checks passed

Frando deleted the Frando/pathstats branch February 9, 2026 08:45

github-project-automation bot moved this from 🏗 In progress to ✅ Done in iroh Feb 9, 2026

		let path_stats = self.path_stats(path_id).unwrap_or_default();
		self.path_stats.remove(&path_id);

Conversation

Frando commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Breaking Changes

Notes & open questions

Uh oh!

github-actions bot commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-commenter commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions bot commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Performance Comparison Report

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

flub left a comment

Choose a reason for hiding this comment

Uh oh!

flub Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

Frando Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

flub Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

Frando Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

Frando Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

flub Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

Frando Feb 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Frando commented Jan 30, 2026 •

edited

Loading

github-actions bot commented Jan 30, 2026 •

edited

Loading

codecov-commenter commented Jan 30, 2026 •

edited

Loading

github-actions bot commented Jan 30, 2026 •

edited

Loading