WildFly can stuck for a while when trying to connect an unavailable Kafka socket and is terminated -------------------------------------------------------------------------------------------------- Key: WFWIP-367 URL: https://issues.redhat.com/browse/WFWIP-367 Project: WildFly WIP Issue Type: Bug Reporter: Jan Stourac Assignee: Kabir Khan Priority: Major Attachments: helloworld-sources.jar, helloworld.war It looks WildFly can stuck for a while with the newly introduced MP Reactive Messaging implementation. If one deploy a deployment that utilizes 'smallrye-kafka' connector it then repeatedly tries to connect against Kafka instance. If that instance is not available on configured socket, then undeploy operation of the app stucks for a while. Similar thing happens for shutdown operation of the server via regular 'ctrl+C' signal. Based on the discussion, this may be an issue that needs to be addressed on SmallRye side as there may be some timeout reached before Kafka connector stops listening. From user point of view - both these operations are not 'graceful' as such - in both cases user expects that undeploy and shutdown should be performed immediately. Thinking about it, I think that even in case of a graceful shutdown, since these are just an attempts for a connection, there is still no need to linger with server shutdown.