feat(transport): Use a shorter ECN probe threshold initially #1964
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit adds a test where a client connects to a server over a connection dropping all ECN marked datagrams (ECN blackhole) in both directions, asserting 43 RTT to detect ECN blackhole, disable ECN and eventually establish connection.
Failed Interop TestsQUIC Interop Runner, client vs. server
All resultsSucceeded Interop TestsQUIC Interop Runner, client vs. server
Unsupported Interop TestsQUIC Interop Runner, client vs. server
|
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1964 +/- ##
=======================================
Coverage 94.96% 94.97%
=======================================
Files 112 112
Lines 36504 36525 +21
=======================================
+ Hits 34667 34689 +22
+ Misses 1837 1836 -1 ☔ View full report in Codecov by Sentry. |
Benchmark resultsPerformance differences relative to e05cc70. coalesce_acked_from_zero 1+1 entries: No change in performance detected. time: [190.00 ns 190.59 ns 191.19 ns]
change: [-0.1604% +0.2995% +0.7300%] (p = 0.20 > 0.05)
coalesce_acked_from_zero 3+1 entries: No change in performance detected. time: [230.40 ns 231.00 ns 231.63 ns]
change: [-0.6668% -0.1846% +0.2545%] (p = 0.44 > 0.05)
coalesce_acked_from_zero 10+1 entries: No change in performance detected. time: [230.59 ns 231.29 ns 232.14 ns]
change: [-1.8009% -0.7863% -0.0234%] (p = 0.08 > 0.05)
coalesce_acked_from_zero 1000+1 entries: Change within noise threshold. time: [212.48 ns 212.78 ns 213.10 ns]
change: [+0.4856% +1.0617% +1.6352%] (p = 0.00 < 0.05)
RxStreamOrderer::inbound_frame(): Change within noise threshold. time: [119.03 ms 119.10 ms 119.17 ms]
change: [+0.3308% +0.4269% +0.5194%] (p = 0.00 < 0.05)
transfer/Run multiple transfers with varying seeds: No change in performance detected. time: [53.900 ms 56.938 ms 60.007 ms]
thrpt: [66.659 MiB/s 70.252 MiB/s 74.212 MiB/s]
change:
time: [-6.1261% +1.5777% +9.3197%] (p = 0.69 > 0.05)
thrpt: [-8.5252% -1.5532% +6.5259%]
transfer/Run multiple transfers with the same seed: No change in performance detected. time: [67.657 ms 73.881 ms 80.075 ms]
thrpt: [49.953 MiB/s 54.141 MiB/s 59.122 MiB/s]
change:
time: [-15.215% -3.5097% +8.3454%] (p = 0.56 > 0.05)
thrpt: [-7.7026% +3.6374% +17.946%]
1-conn/1-100mb-resp (aka. Download)/client: No change in performance detected. time: [280.38 ms 285.37 ms 290.13 ms]
thrpt: [344.67 MiB/s 350.43 MiB/s 356.66 MiB/s]
change:
time: [-1.7285% +0.6221% +2.9358%] (p = 0.62 > 0.05)
thrpt: [-2.8520% -0.6183% +1.7589%]
1-conn/10_000-parallel-1b-resp (aka. RPS)/client: No change in performance detected. time: [443.92 ms 447.68 ms 451.48 ms]
thrpt: [22.149 Kelem/s 22.337 Kelem/s 22.527 Kelem/s]
change:
time: [-1.7177% -0.6391% +0.4003%] (p = 0.26 > 0.05)
thrpt: [-0.3987% +0.6432% +1.7478%]
1-conn/1-1b-resp (aka. HPS)/client: No change in performance detected. time: [68.561 ms 68.850 ms 69.183 ms]
thrpt: [14.454 elem/s 14.524 elem/s 14.586 elem/s]
change:
time: [-0.9345% -0.1946% +0.5412%] (p = 0.61 > 0.05)
thrpt: [-0.5383% +0.1950% +0.9433%]
Client/server transfer resultsTransfer of 33554432 bytes over loopback.
|
|
Friendly ping @larseggert. |
|
Thanks. I'll add a fix so that we stop trying ECN earlier in this case. |
larseggert
reviewed
Jul 15, 2024
|
@mxinden WDYT? |
mxinden
commented
Jul 15, 2024
Co-authored-by: Max Inden <mail@max-inden.de> Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Max Inden <mail@max-inden.de> Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Max Inden <mail@max-inden.de> Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Max Inden <mail@max-inden.de> Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Max Inden <mail@max-inden.de> Signed-off-by: Lars Eggert <lars@eggert.org>
mxinden
commented
Jul 15, 2024
There was a problem hiding this comment.
@larseggert I cleaned-up the tests. Mind taking another look?
|
@KershawChang or @martinthomson, could you take a look? This would let us enable ECN in Gecko w/@mxinden's patch. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit adds a test where a client connects to a server over a connection dropping all ECN marked datagrams (ECN blackhole) in both directions, asserting 43 RTT to detect ECN blackhole, disable ECN and eventually establish connection. Test assumes 20 RTT for client to detect blackhole, 20 RTT for server to detect blackhole and 3 RTT for handshake to be confirmed.
Test for #1925.
How likely are ECN blackholes on today's Internet? Should this block ECN roll-out on Firefox? //CC @larseggert
Bauer et. alia. in "Measuring the State of ECN Readiness in Servers, Clients, and Routers" suggests that ECN blackholes are rare.
Test output