When using a high available setup, an issue occurs when sending a message with a header that is too large. The main broker rejects the message, but at the same time, it seems that it is trying to perform replication to the backup broker.

The backup broker also rejects this replication message, but after this it seems that all further messages are blocked. Only after killing the backup broker, the messages will be processed. Seems like this is related to journal related operations being blocked until backup acknowledges the completion of synchronization (https://activemq.apache.org/components/artemis/documentation/latest/ha.html#replication)

I have included a Quarkus demo app with a description of the scenario, and my main and backup broker configurations to replicate the issue. Tested against Artemis version 2.38

Issue sounds similar to issue discussed in ARTEMIS-4694