[jboss-jira] [JBoss JIRA] Commented: (JBAS-6144) Resolve intermittent org.jboss.test.cluster.defaultcfg.web.test.ScopedSetAttributeTestCase(Default-udp).testInvalidate failure

[Brian Stansberry (JIRA)]

    [ https://jira.jboss.org/browse/JBAS-6144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12535983#action_12535983 ] 

Brian Stansberry commented on JBAS-6144:
----------------------------------------

This test later gets re-run using REPL_SYNC. I don't want to move many tests to purely REPL_SYNC, no REPL_ASYNC coverage because our standard config is REPL_ASYNC and I think we need to test things that way I would love it if our test infrastructure could support async replication but without sleeps. JBC testsuite had a mechanism for doing that (involved listeners to track when replication messages arrived) but that was based on all nodes in the cluster + the test driver running in the same VM. AS clustering testsuite has some tests that work that way (in o.j.test.cluster.defaultcfg.simpleweb.test package) but most are true integration tests where there are 2 AS processes and a remove test driver.

The above comment is about the general issue of how to make the clustering testsuite more robust. The specific issue here is this particular test method seemed to fail more often. TBH, I'm not sure if that is still the case; this JIRA might be out of date. Before spending much effort chasing this particular failure, it's probably worthwhile to track the testsuite results for a while and see if it still pops up.

> Resolve intermittent org.jboss.test.cluster.defaultcfg.web.test.ScopedSetAttributeTestCase(Default-udp).testInvalidate failure
> ------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: JBAS-6144
>                 URL: https://jira.jboss.org/browse/JBAS-6144
>             Project: JBoss Application Server
>          Issue Type: Task
>      Security Level: Public(Everyone can see) 
>          Components: Clustering, Web (Tomcat) service
>            Reporter: Brian Stansberry
>            Assignee: Brian Stansberry
>             Fix For: 6.0.0.Final
>
>
> This test intermittently fails.  The web clustering tests in the default REPL_ASYNC config can occasionally fail due to timing, where the test fails over to the other cluster node before the replicated session arrives.  There is a 200 ms delay before failover, but occasional bad luck (e.g. a full gc at wrong time) can make that too short.  (Making it longer makes the whole testsuite longer.)  But, this issue should cause random failures, whereas this particular test seems to pop up as a failure a high percentage of the time. Need to understand why.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira