When amqpMinLargeMessageSize is configured in a broker cluster, messages still goes to the large message folder, and not getting clean up until a restart of the broker.

This causes the storage to be filled up. 

The root of the problem is that there are inconsistent places to configure MinLargeMessageSize, and even amqp protocol treats a 200k message as non large message, when the message is forwarded to another broker, the core bridge does not detect this is a large message and causing the messages not getting clean up. 

1. Large message handling are configuring in various places: core and openwire can only be configured at the client level. While AMQP and Stomp are configured at the Acceptor part. [1]
2. There is no instructions on how to configure that in a cluster, and there was an issue when amqpMinLargeMessageSize

Large messages configuration should have been handled by a single configuration.

We should provide examples of how to configure minLargeMessageSize in a broker cluster. 

 [1]

Core: tcp://localhost:61616?minLargeMessageSize=250000

Openwire: tcp://localhost:61616?wireFormat.minLargeMessageSize=250000

AMQP: amqpMinLargeMessageSize=204800</acceptor>

{{Stomp: Acceptor }}stompMinLargeMessageSize=204800{{ }}