Currently ActiveMQ Artemis only restricts normal messages from paging at the address level using page-limit-bytes, and whenever any address exceeds this limit we are able to block producers from sending messages to the address by setting the page-full-policy to FAIL.
However, this page-limit-bytes does not cover the large-messages directory in the data directory due to which there is no track for large messages. If any consumer for a queue is down for sometime for which large messages are being sent, the storage is getting full impacting other customers hosted in the broker. Ideally it should only impact the particular tenant.
We have max-disk-usage in place, but it's at global level and impacting other customers if disk usage is beyond this limit. Could you please check if there is any parameter to restrict large messages. We tested and seems the destination parameter "Persistent Size" is the sum of the disk used by paging, large messages, journals and bindings, but I don't see any restriction based on this parameter.
There is currently no setting to limit disk usage on a per-address level. I've opened ARTEMIS-4700 to implement this feature.