]
Brian Stansberry updated WFLY-6849:
-----------------------------------
Component/s: JMS
Duplicate messages in replicated HA topology when backup is
shutdowned
----------------------------------------------------------------------
Key: WFLY-6849
URL:
https://issues.jboss.org/browse/WFLY-6849
Project: WildFly
Issue Type: Bug
Components: JMS
Reporter: Bartosz Baranowski
Assignee: Clebert Suconic
Priority: Critical
Labels: downstream_dependency
Scenario
# Configure 2 nodes in replicated dedicated topology
# Start live (node-1) and backup (node-2)
# Start producer
# Shut down node-2
# Stop producer
# Check if there are some duplicates on node-1 using CLI operation list-messages
Expectation: there is no duplications
Actual state: there are 10 messages with the same _AMQ_DUPL_ID twice
After that Backup is shut downed, the Live is not able to replicate its data to Backup.
It waits 30 seconds until timeouts expire. Meanwhile producer does not get response from
Live and it gets TimeoutException on commit. It retries to send the same batch of messages
and then commit them again. I think that problem is at this point. The Live does not
detect duplicate messages.