AOP - Using Retryable and Recoverable with a suspend function fails

[tmapes]

Expected Behavior

A suspend function annotated with both @Retryable and @Recoverable should retry as many times as configured in the Retryable annotation before triggering the Recoverable Fallback.

Actual Behaviour

During execution, a ClassCastException is raised during the RecoveryInterceptor that causes the Fallback to be executed during execution of the declared function.

This causes the calling function to get the 'Fallback' output before the retries are completed.

This also leads to the retries happening in the background, despite the fact that the Fallback has already fired.

The below stacktrace is logged:

java.lang.ClassCastException: class kotlin.Unit cannot be cast to class java.util.concurrent.CompletableFuture (kotlin.Unit is in unnamed module of loader 'app'; java.util.concurrent.CompletableFuture is in module java.base of loader 'bootstrap')
	at io.micronaut.retry.intercept.RecoveryInterceptor.lambda$fallbackForFuture$3(RecoveryInterceptor.java:167)
	at java.base/java.util.concurrent.CompletableFuture.uniWhenComplete(CompletableFuture.java:859)
	at java.base/java.util.concurrent.CompletableFuture$UniWhenComplete.tryFire(CompletableFuture.java:837)
	at java.base/java.util.concurrent.CompletableFuture.postComplete(CompletableFuture.java:506)
	at java.base/java.util.concurrent.CompletableFuture.completeExceptionally(CompletableFuture.java:2088)
	at io.micronaut.aop.util.CompletableFutureContinuation.resumeWith(CompletableFutureContinuation.kt:51)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:46)
	at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:106)
	at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:571)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:750)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:678)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:665)

Steps To Reproduce

• Declare a suspend function and annotate it with both Recoverable and Retryable.
• Implement a sufficient Fallback implementation for this function
• Throw an exception inside the suspend function that should trigger a retry
• The callers invocation of this suspend function is then 'completed' with the Fallback's return value while retries continue in the background.
  • My example repo has the suspend function return Unit however the same issue happens when you return a value, I tested with Boolean

Environment Information

OS - macOS 12.2.1 (x86_64)
JVM - 11.0.11

Example Application

https://github.com/tmapes/micronaut-suspend-retry-recover

Version

3.3.4

[hollingsworthd]

Thanks for the bug report and branch to reproduce this.

I reproduced this bug and investigated, but I am not a Kotlin expert -- @sdelamo should someone else take this on? Un-assigning myself for the moment.

[tmapes]

Confirmed fixed in https://github.com/micronaut-projects/micronaut-core/releases/tag/v3.8.6, thanks folks!