I am running Artemis with the default config. The config includes <max-size-bytes>10485760</max-size-bytes> and <address-full-policy>BLOCK</address-full-policy> for every address. I used simple_send.py AMQP example client from qpid-proton to send messages onto a jms.queue.myqueuembm queue.

I observed that the broker continues to receive messages until the Java process it runs out of RAM, not until it exhausts the queue size limit.

I tried sending both large, 1MiB messages, and small messages with nothing or almost nothing in them. What I described above happened in both cases. Router accepted 300 large messages and several hundreds of thousands of small messages, before running out of memory. Then it became unresponsive. In case of small messages, I had to kill -9 it afterwards, nothing else worked.