enable Netty native transports by default
enable Netty native transports (epoll, kqueue) by default
Codecov Report
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 61.94%. Comparing base (
59551e4) to head (e6f158b). Report is 888 commits behind head on master.
Additional details and impacted files
@@ Coverage Diff @@
## master #12577 +/- ##
============================================
+ Coverage 61.75% 61.94% +0.19%
+ Complexity 207 198 -9
============================================
Files 2436 2558 +122
Lines 133233 141208 +7975
Branches 20636 21974 +1338
============================================
+ Hits 82274 87478 +5204
- Misses 44911 47069 +2158
- Partials 6048 6661 +613
| Flag | Coverage Δ | |
|---|---|---|
| custom-integration1 | <0.01% <0.00%> (-0.01%) |
:arrow_down: |
| integration | <0.01% <0.00%> (-0.01%) |
:arrow_down: |
| integration1 | <0.01% <0.00%> (-0.01%) |
:arrow_down: |
| integration2 | 0.00% <0.00%> (ø) |
|
| java-11 | 61.89% <100.00%> (+0.19%) |
:arrow_up: |
| java-21 | 61.84% <100.00%> (+0.21%) |
:arrow_up: |
| skip-bytebuffers-false | 61.91% <100.00%> (+0.16%) |
:arrow_up: |
| skip-bytebuffers-true | 61.82% <100.00%> (+34.09%) |
:arrow_up: |
| temurin | 61.94% <100.00%> (+0.19%) |
:arrow_up: |
| unittests | 61.94% <100.00%> (+0.19%) |
:arrow_up: |
| unittests1 | 46.38% <100.00%> (-0.51%) |
:arrow_down: |
| unittests2 | 27.81% <0.00%> (+0.07%) |
:arrow_up: |
Flags with carried forward coverage won't be shown. Click here to find out more.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Correct me if I am wrong. There is unlikely to be any negative impact for this. In fact this will only lead to may be P99 improvements at high throughput ?
cc @jasperjiaguo
Correct me if I am wrong. There is unlikely to be any negative impact for this. In fact this will only lead to may be P99 improvements at high throughput ?
Yes we've only seen benefit in high qps usecases (as expected).
@sullis may I know what's the rational behind this change? Did you see improvements?
Correct me if I am wrong. There is unlikely to be any negative impact for this. In fact this will only lead to may be P99 improvements at high throughput ?
Yes we've only seen benefit in high qps usecases (as expected).
@sullis may I know what's the rational behind this change? Did you see improvements?
I am new to the Apache Pinot ecosystem. My previous experience with Netty was operating Netflix Zuul. Zuul uses Netty with epoll on Linux.
This PR is a draft because it needs further discussion. I noticed that there is an analysis of epoll in an earlier PR: https://github.com/apache/pinot/pull/8300