!test Migrate akka-multi-node-testkit to Netty4

[He-Pin]

References #31974

Not sort the imports yet.

[johanandren]

Huh, something weird with validation, I'll close and re-open to retrigger.

[He-Pin]

@johanandren I think this is ready.

[patriknw]

Looking good, thanks.

Note that we don't run multi-jvm tests in PR validation, so before merging we should verify locally.

[johanandren]

I think we can also trigger the multi-jvm tests manually against this branch before merging.

[johanandren]

Seeing failures with this exception when I run it locally:

 WARNING: Failed to initialize a channel. Closing: [id: 0xe92cb4cd, L:/127.0.0.1:4711 - R:/127.0.0.1:54387]
[error] [JVM-1-LeaderElectionWithFailureDetectorPuppetMultiJvmNode1] io.netty.channel.ChannelPipelineException: akka.remote.testconductor.ConductorHandler is not a @Sharable handler, so can't be added or removed multiple times.
[error] [JVM-1-LeaderElectionWithFailureDetectorPuppetMultiJvmNode1] 	at io.netty.channel.DefaultChannelPipeline.checkMultiplicity(DefaultChannelPipeline.java:600)
[error] [JVM-1-LeaderElectionWithFailureDetectorPuppetMultiJvmNode1] 	at io.netty.channel.DefaultChannelPipeline.addLast(DefaultChannelPipeline.java:202)
[error] [JVM-1-LeaderElectionWithFailureDetectorPuppetMultiJvmNode1] 	at io.netty.channel.DefaultChannelPipeline.addLast(DefaultChannelPipeline.java:195)
[error] [JVM-1-LeaderElectionWithFailureDetectorPuppetMultiJvmNode1] 	at akka.remote.testconductor.TestConductorPipelineFactory.$anonfun$initChannel$1(RemoteConnection.scala:81)
...

[He-Pin]

@johanandren I have updated, and add the @Shareable annotation. have very bad network connection, can you check it out again?

[johanandren]

Now the multi jvm cluster tests are failing with a different error, connection failed, did you get those to pass locally with these changes? akka-cluster/MultiJvm/test

[He-Pin]

Let me check it then.

[johanandren]

Still seeing tests failing with connection refused errors when I try to run cluster multi JVM tests with this

[He-Pin]

Still seeing tests failing with connection refused errors when I try to run cluster multi JVM tests with this

hum，let's
me check it this weekend.

[johanandren]

No urgency, vacation times in Europe, so reviewing will be slow the next couple of weeks anyway.

[johanandren]

LGTM, ran through cluster multi jvm tests on both MacOS and Linux with all tests green with these changes. 👍

Thanks @He-Pin

[He-Pin]

@johanandren Yes, it did take me sometime to make it right, and now, Netty 3 is gone.

[patriknw]

is slf4j/logback needed by Netty or what is the reason we need this now?

is this included by the other sbt projects that have multi-jvm tests or is same needed in more places?

[patriknw]

I see this in logs:

2023-08-14T08:44:19.0726537Z �[0m[�[0m�[31merror�[0m] �[0m�[0m[�[34mJVM-2�[0m] SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".�[0m
2023-08-14T08:44:19.0727622Z �[0m[�[0m�[31merror�[0m] �[0m�[0m[�[34mJVM-2�[0m] SLF4J: Defaulting to no-operation (NOP) logger implementation�[0m

[He-Pin]

I need it to suppress netty 4's logging

[johanandren]

I'll add it for all multi-jvm tests then.

[patriknw]

I have tried this in CI: https://github.com/akka/akka/actions/runs/5853551428
Seems to work fine, we have some flaky tests since before.

[patriknw]

I see this in logs:

2023-08-14T08:44:19.0726537Z �[0m[�[0m�[31merror�[0m] �[0m�[0m[�[34mJVM-2�[0m] SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".�[0m
2023-08-14T08:44:19.0727622Z �[0m[�[0m�[31merror�[0m] �[0m�[0m[�[34mJVM-2�[0m] SLF4J: Defaulting to no-operation (NOP) logger implementation�[0m

[patriknw]

I created issue #32038 but that happened once before this PR as well.