Hi!!! AFAIK the en-queued persistent messages are in the journal and in the heap too until consumed, but if Artemis should run low on memory it could page them on disk (if a paging policy is configured). Just to ask: why let the broker seems to behave like a database for the error messages? IMHO an efficient use of the broker is to let it be like a dispatcher/router of messages and not as a final destination for them. Consuming/moving them and using a proper tool (time series database or others) to collect/query the error messages or rise alarm of any kind would be more sympathetic with the broker mechanics and purpose. Or I've misunderstood the way you're using the broker for the error queue?
P.S: there is a nice article http://sensatic.net/messaging/messaging-anti-patterns-part-1.html <http://sensatic.net/messaging/messaging-anti-patterns-part-1.html> on it too! -- View this message in context: http://activemq.2283324.n4.nabble.com/How-does-Apache-Artemis-manage-heap-space-RAM-tp4723220p4723308.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
