Francesco Nigro

@vietj It seems I cannot make the io_uring part to complete because of intermittent epoll failures :O no idea what's going on

This is the list of errors I'm getting by running separately the IO uring tests with the github CI: ``` Tests in error: DatagramTest.testPauseResume:209->AsyncTestBase.await:121->AsyncTestBase.await:133 » IllegalState Http1xTest>AsyncTestBase.after:83->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState Http2MetricsTest>AsyncTestBase.after:83->HttpMetricsTestBase.tearDown:57->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183...

New results after changing the UDP test: ``` Results : Tests in error: Http1xTest>AsyncTestBase.after:83->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState Http2MetricsTest>AsyncTestBase.after:83->HttpMetricsTestBase.tearDown:57->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState Http2MetricsTest>AsyncTestBase.after:83->HttpMetricsTestBase.tearDown:57->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState Http2Test>AsyncTestBase.after:83->HttpTestBase.tearDown:71->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState NetTest>AsyncTestBase.after:83->tearDown:132->VertxTestBase.tearDown:87->AsyncTestBase.close:662->AsyncTestBase.awaitLatch:598->AsyncTestBase.awaitLatch:602->AsyncTestBase.assertTrue:380->AsyncTestBase.handleThrowable:183 » IllegalState Tests run:...

Hi @vietj Here https://github.com/eclipse-vertx/vert.x/pull/4164#issuecomment-1141100555 I am just come back from the summer holidays - will be back on my desk on Monday 👍

@tsegismont See https://bugs.openjdk.java.net/browse/JDK-8161588?focusedCommentId=13974951&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-13974951 for more info The TLDR story is that some method handle (missing field) lookups using lambda can stealthy throws exceptions: so, it won't affect hot code (because...

Current logic is not enforcing an issued resume to NOT be emitted again regardless the state of any previously submitted one - that means that a "lucky" concurrent submit can...

@vietj I've changed the benchmark and something that requires investigation has happened: ``` Benchmark (actionWork) (type) Mode Cnt Score Error Units CombinerExecutorBenchmark.submitFourThreads 0 sync thrpt 20 23.363 ± 1.548 ops/us...

> I think we are not looking to compare this to the semaphore version because we are not comparing apples to apples. The semaphore version will create contention and potentially...

These are the numbers with the `master` benchmark (using 2 threads): ``` Benchmark Mode Cnt Score Error Units CombinerExecutorBenchmark.impl thrpt 10 11484.804 ± 5100.280 ops/ms ``` These are by using...

 This is screen from JFR allocation recording of Techempower's plaintext, showing this as one of the top allocators