[jboss-jira] [JBoss JIRA] (WFLY-10607) Fix intermittent failures in RemoteEJBForwardingTestCases

[Richard Achmatowicz (JIRA)]

    [ https://issues.jboss.org/browse/WFLY-10607?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13610644#comment-13610644 ] 

Richard Achmatowicz edited comment on WFLY-10607 at 7/26/18 8:33 AM:
---------------------------------------------------------------------

I have found out why the test was still failing. Upon looking at the test environment setup, it was observed that the two clusters 
{code}
ejb-forwarder = {node-1, node-2} 
{code} and 
{code}
ejb={node-3, node4} 
{code}
were not forming correctly.  Once the cluster setup was fixed, and with the above mentioned change to the synchronization interceptor, the test now passes.

What was the problem with the test case setup and why were the clusters not forming correctly? The test case setup depends on several steps carried out in the testsuite/integration/clustering pom.xml file:
* use a wildfly-maven-plugin execution to adjust the stack of the base distribution used in the test via clustering-all.cli
* use a maven antrun-plugin execution to copy the distribution to the four server distributions wildfly-1, wildfly-2, wildfly-3, wildfly4
* use a maven-surefilre-plugin execution to run the 2 clusters test using the four servers which have been started with the default socket-binding multicast addresses  (this by default causes the four nodes to all be members of the same cluster when started)
* use a ServerSetupTask in the test case itself to further configure the four nodes so that they form  two separate clusters suitable for the test

It was noticed that the two clusters were not forming correctly. Upon further investigation using the management console, it was seen that the multicast addresses were correctly configured in each server's management interface, but that the actual multicast addresses used were not correct. 

This was traced to the following issue: https://issues.jboss.org/browse/WFCORE-3977. What was happening was that the ServerSetupTask was correctly making the changes via the management CLI, but because the servers were not put into reload-required, the changes were not taking effect and the clusters were not forming correctly. Forcing a reload of the servers causes the servers to configure correctly. 

With correctly configured server environment, plus the changes to the synchronization interceptor mentioned above, the tests now pass.





was (Author: rachmato):
I have found out why the test was still failing. Upon looking at the test environment setup, it was observed that the two clusters ejb-forwarder = {node-1, node-2} and ejb={node-3, node4} were not forming correctly.  Once the cluster setup was fixed, and with the above mentioned change to the synchronization interceptor, the test now passes.

What was the problem with the test case setup and why were the clusters not forming correctly? The test case setup depends on several steps carried out in the testsuite/integration/clustering pom.xml file:
* use a wildfly-maven-plugin execution to adjust the stack of the base distribution used in the test via clustering-all.cli
* use a maven antrun-plugin execution to copy the distribution to the four server distributions wildfly-1, wildfly-2, wildfly-3, wildfly4
* use a maven-surefilre-plugin execution to run the 2 clusters test using the four servers which have been started with the default socket-binding multicast addresses  (this by default causes the four nodes to all be members of the same cluster when started)
* use a ServerSetupTask in the test case itself to further configure the four nodes so that they form  two sepaate clusters suitable for the test

It was noticed that the two clusters were not forming correctly. Upon further investigation using the management console, it was seen that the multicast addresses were correctly configured in each server's management interface, but that the actual multicast addresses used were not correct. 

This was traced to the following issue: https://issues.jboss.org/browse/WFCORE-3977. What was happening was that the ServerSetupTask was correctly making the changes via the management CLI, but because the servers were not put into reload-required, the changes were not taking effect and the clusters were not forming correct. Forcing a reload of the servers causes the servers to configure correctly. 

With correctly configured server environment, pluse the changes to the synchronization interceptor mentioned above, the tests now pass.




> Fix intermittent failures in RemoteEJBForwardingTestCases
> ---------------------------------------------------------
>
>                 Key: WFLY-10607
>                 URL: https://issues.jboss.org/browse/WFLY-10607
>             Project: WildFly
>          Issue Type: Bug
>          Components: EJB, Test Suite
>            Reporter: Petr Kremensky
>            Assignee: Richard Achmatowicz
>
> The tests were ignored by https://github.com/wildfly/wildfly/pull/10563/files



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)