[ https://issues.apache.org/jira/browse/ARTEMIS-2709?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17087351#comment-17087351 ]
Atri Sharma commented on ARTEMIS-2709: -------------------------------------- Some naive comments here: What I see happening is that post the ingestion of a message ( {code:java} ClientConsumerImpl::handleRegularMessage {code} ), we queue a runnable which will invoke {code:java} ClientConsumerImpl::callOnMessage {code} which in turn will poll the message from the underlying priority queue buffer, thus removing it from the buffer. Since every ingestion has a runnable queued, that effectively means that all messages that are brought into the system are polled from the buffer even before scaleDownDelay gets to call recieveDurableMessages, which will seek for messages on the buffer. Hence, receiveDurableMessages sees an empty buffer and the assertion for non null messages fails. What am I missing? > Fix LiveToLiveFailoverTest::scaleDownDelay > ------------------------------------------- > > Key: ARTEMIS-2709 > URL: https://issues.apache.org/jira/browse/ARTEMIS-2709 > Project: ActiveMQ Artemis > Issue Type: Task > Reporter: Clebert Suconic > Priority: Major > Fix For: 2.12.0 > > -- This message was sent by Atlassian Jira (v8.3.4#803005)