Quarkus caches Failures

[wesleysalimansdvb]

Describe the bug

Whenever a failure occurs such as a Timeout this failure is cached, so each subsequent call will retrieve this failure from cache (and the remote service is not called, until the cache is invalidated).

Expected behavior

Quarkus does not cache the failure.

Actual behavior

Quarkus caches the failure and each subsequent call for that cache key get the failure from cache (the remote service is not called)

How to Reproduce?

@CacheResult(cacheName = "username")
  public Uni<String> getUsernameById(String userId) {
    return remoteApi
        .getUser(userId)
        .onFailure()
        .invoke(failure -> Log.error("Error when getUsername", failure))
        .onItem()
        .transform(User::getName);
  }

Output of uname -a or ver

Darwin REDACTED 23.3.0 Darwin Kernel Version 23.3.0: Wed Dec 20 21:30:44 PST 2023; root:xnu-10002.81.5~7/RELEASE_ARM64_T6000 arm64

Output of java -version

openjdk 21.0.1 2023-10-17 LTS OpenJDK Runtime Environment Corretto-21.0.1.12.1 (build 21.0.1+12-LTS) OpenJDK 64-Bit Server VM Corretto-21.0.1.12.1 (build 21.0.1+12-LTS, mixed mode, sharing)

Quarkus version or git rev

3.9.0.CR1

Build tool (ie. output of mvnw --version or gradlew --version)

maven

Additional information

No response

[quarkus-bot]

/cc @gwenneg (cache)

[gsmet]

@gwenneg could you have a look?

Now I'm not entirely sure that's something we can fix without introducing an annotation to specify the behavior as caching the failure could be a valid behavior too.

[stefvanderweldevolksbanknl]

@gwenneg could you have a look?

Now I'm not entirely sure that's something we can fix without introducing an annotation to specify the behavior as caching the failure could be a valid behavior too.

It doesn't seem logical to me that failures are cached, but Exceptions are not.

[wesleysalimansdvb]

And if this is expected behaviour, what would be the recommended way to ensure that subsequent calls of a failed request do call the remote service and don't get the cached failure?

[geoand]

I consider this behavior incorrect and have opened #39762 to address it.