[
https://jira.jboss.org/jira/browse/JBMESSAGING-1722?page=com.atlassian.ji...
]
Howard Gao updated JBMESSAGING-1722:
------------------------------------
Attachment: test-1722.zip
The zip file contains the test case. It runs under jbm_install/examples dir.
Run ant to print the test steps.
Brief description:
1. setup a cluster with two nodes, node 0 and node 1.
2. setup a clustered queue myTestQ with full size 8000, pageSize/downcacheSize 2000.
3. send 18000 messages
to myTestQ on node 0
4. receive 4000 messages from myTestQ on node 0
5. receive 72 messages
6. receive 2000 messages in a tx, but roll back it.
7. crach node 1, that will cause the paging false and reset firstPagingOrder and
nextPagingOrder.
8. receive 1 message, make the queue memory list one message less than full size.
9. send 4061 messages to make paging happen again
10. start and crash node 1 again, make page flushed and disable paging.
11. receive 6000 messages.
12. restart node0 and see the exceptions.
MessagingQueue.mergeIn() can break the paging order.
----------------------------------------------------
Key: JBMESSAGING-1722
URL:
https://jira.jboss.org/jira/browse/JBMESSAGING-1722
Project: JBoss Messaging
Issue Type: Bug
Affects Versions: 1.4.0.SP3.CP08, 1.4.4.GA
Reporter: Howard Gao
Assignee: Howard Gao
Fix For: 1.4.0.SP3.CP09, 1.4.5.GA
Attachments: test-1722.zip
On node failover, clustered queue on the failed node will be merged into the queue on the
back up node. If the local queue is in paging state but the failed queue has no messages
to merge in, it will reset the firstPagingOrder and nextPagingOrder and therefore change
the paging state to false. That makes the paging broken.
This issue were found during investigating the associated ticket.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira