I just got a pretty interesting observation: I just tried my own tweaked PendingDurableSubscriberPolicy (as I learned from the other thread, see below) to the test code I've been using:
http://activemq.2283324.n4.nabble.com/How-to-configure-this-combination-of-queue-cursors-tt4670003.html#a4670027 Then, the test code went through: MemoryLimitTest.java <http://activemq.2283324.n4.nabble.com/file/n4670029/MemoryLimitTest.java> ----- Let's explain what's happened again in the test code above, shortly. With only durable subscriber with prefetchSize=0, persistent message publishing to the topic is blocked (PFC) by hitting per-destination memory limit (1MB). But, when I changed the PendingDurableSubscriberPolicy from the default "StorePendingDurableSubscriberMessageStoragePolicy" to the customized one (see below), persistent message publishing didn't get stuck by any PFC but went through very well. // This makes the test flow through! // Without this (with default StorePendingDurableSubscriberMessageStoragePolicy), // persistent message publishing to the topic got stuck. Why?? policyEntry.setPendingDurableSubscriberPolicy(new StorePendingDurableSubscriberMessageStoragePolicy() { @Override public PendingMessageCursor getSubscriberPendingMessageCursor(Broker broker, String clientId, String name, int maxBatchSize, DurableTopicSubscription sub) { broker.getBrokerService().setPersistent(false); StoreDurableSubscriberCursor cursor = (StoreDurableSubscriberCursor) super .getSubscriberPendingMessageCursor(broker, clientId, name, maxBatchSize, sub); broker.getBrokerService().setPersistent(true); return cursor; } }); I'm curious what difference is there between these two "PendingDurableSubscriberPolicy"s particularly for the persistent pending cursor used inside.. Strange.. -- View this message in context: http://activemq.2283324.n4.nabble.com/cursor-memory-usage-limit-vs-memory-usage-limit-tp4669679p4670029.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
