Fix blocking annotation not working on gRPC calls properly

[Dogacel]

Motivation:

#5295

Modifications:

• Add an explicit property on ctx to determine whether the blocking task executor should be used or not.

Result:

• Closes #5295

Maybe we could have just checked whether the blockingTaskExecutor is null or not but I did not like that idea. Also this is the recommended way in the issue description.

Will add tests soon.

[github-actions[bot]]

🔍 Build Scan® (commit: d458ec4bc42400db37f0ff17b6805a79efe2eecd)

Job name	Status	Build Scan®
build-self-hosted-unsafe-jdk-8	✅	https://ge.armeria.dev/s/ghkneexuz56qe
build-self-hosted-unsafe-jdk-21-snapshot-blockhound	❌ (failure)	https://ge.armeria.dev/s/rn5ygxteecvvo
build-self-hosted-unsafe-jdk-17-min-java-17-coverage	✅	https://ge.armeria.dev/s/qnj47ptboc5ng
build-self-hosted-unsafe-jdk-17-min-java-11	✅	https://ge.armeria.dev/s/y2so6uqgmbuq4
build-self-hosted-unsafe-jdk-17-leak	❌ (failure)	https://ge.armeria.dev/s/vdcoy5z26clf6
build-self-hosted-unsafe-jdk-11	✅	https://ge.armeria.dev/s/sgqven4bsl2ta
build-macos-12-jdk-21	❌ (failure)	https://ge.armeria.dev/s/syvwdy6btucwu

[trustin]

The current plan is to deprecate ctx.eventLoop and provide a way for a user to specify an arbitrary executor or even a virtual thread executor via ServiceConfig.serviceWorkerGroup, which will eventually make blockingTaskExecutor obsolete.

[Dogacel]

The current plan is to deprecate ctx.eventLoop and provide a way for a user to specify an arbitrary executor or even a virtual thread executor via ServiceConfig.serviceWorkerGroup, which will eventually make blockingTaskExecutor obsolete.

Do you prefer to close this PR and continue with the long term approach or resolve this issue first and iteratively move towards the ultimate goal of removing blocking executor?

[trustin]

Both are fine by me. Let me know if you need this PR to be merged, we'd be happy to accept it given it will take quite some time until we fix this properly. :sweat_smile:

[Dogacel]

Both are fine by me. Let me know if you need this PR to be merged, we'd be happy to accept it given it will take quite some time until we fix this properly. 😅

I think this is a pretty important issue for teams with less experience. Previously, @Blocking was being ignored and service was still using the event loop. We have faced performance issues regarding blocking calls so this bug might exist in many people's codebase and secretly eat out some performance. We might be affected by this as well!

I have went ahead and added tests to verify this fix works PTAL.

[Dogacel]

Bump

[ikhoon]

Checked CI results:

• 3 of them failed by flaky tests.
• The leak will be fixed by #5858