When using the @ConcurrencyLimit annotation added in 7.0, the underlying ConcurrencyThrottleInterceptor blocks (virtual) threads for a potentially very long time. In typical web scenarios, clients may no longer be interested in the result when the blocked methods finally execute, but they execute anyway, wasting resources.

As a user, I would expect this behavior to be documented in the @ConcurrencyLimit Javadoc (note that the way the Lock is used, there is no guarantee of fairness). Ideally, I would be able to specify a different behavior on the annotation (e.g. throw an exception that I can map to a 429 HTTP status code).

Thank you!