[ 
https://issues.apache.org/jira/browse/IGNITE-28355?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Alexandr Shapkin updated IGNITE-28355:
--------------------------------------
    Labels: MakeTeamcityGreenAgain ignite-3  (was: ignite-3)

> ItNodeTest.testNewPeersConfigurationAppliedListener is potentially flaky due 
> to race between async listener callback and verify
> -------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: IGNITE-28355
>                 URL: https://issues.apache.org/jira/browse/IGNITE-28355
>             Project: Ignite
>          Issue Type: Improvement
>          Components: raft ai3
>            Reporter: Ivan Zlenko
>            Assignee: Ivan Zlenko
>            Priority: Major
>              Labels: MakeTeamcityGreenAgain, ignite-3
>
> The test calls changePeersAndLearnersAsync in a loop, each iteration changing 
> the RAFT group to a single new peer + learner. After each change, it waits 
> for the new leader to appear via waitForCondition, then immediately verifies 
> that onNewPeersConfigurationApplied was called on the mock listener.
>                                                                               
>                                                                               
>                                                                      
> The problem is that onNewPeersConfigurationApplied is scheduled 
> asynchronously on the old leader's common executor, while the test's 
> waitForCondition only checks that the new leader has been elected. The new 
> leader election can complete before the old leader's executor processes the 
> listener callback, causing verify(times(1)) to fail.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to