We have a few microservices implemented using the rx-ified vert.x api (no vert.x callback/future usage), vert.x web for REST, eventbus SPIs with substitutable verticles as different providers, clustered eventbus messaging, and 100% gradle. Implementation, including utility libs, is RxJava 2 reactive.

Considering migrating this to Quarkus, and think I’ve found guide material to help get through a lot of the foundational questions, but still have this one:

Much of the marketing for quarkus emphasizes fast startup times. In our experience, with the clustered eventbus using hazelcast, we don’t even get to app initialization until hazelcast gets the node joined to the cluster, and this process can easily take up to 30 seconds, which is of course considerably longer than the sub-second startup times touted in the quarkus marketing articles.

Does anybody here have any experience bootstrapping a quarkus app that uses a clustered eventbus with hazelcast? Is there any measurable difference in average startup times?