[GitHub] [helix] kaisun2000 edited a comment on issue #1630: Flaky test TestCrushAutoRebalanceNonRack. testLackEnoughInstances

[GitBox <gi...@apache.org>]

kaisun2000 edited a comment on issue #1630:
URL: https://github.com/apache/helix/issues/1630#issuecomment-773106700


   It seems that Huizhi fixed similar issue April 2020 by adding 
   ```
         Assert.assertTrue(TestHelper.verify(() -> {
           _gSetupTool.getClusterManagementTool()
               .enableInstance(CLUSTER_NAME, p.getInstanceName(), false);
           return !InstanceValidationUtil.isEnabled(helixDataAccessor, p.getInstanceName())
               && !InstanceValidationUtil.isAlive(helixDataAccessor, p.getInstanceName());
         }, TestHelper.WAIT_DURATION), "Instance should be disabled and offline");
   ```
   
   This addressed the previous similar issue. There is still existing race condition though. The reason is that from test code perspective at the time this instance is disabled and liveinstance gone, it may not reflect the view of controller. Controller may very well see it the instance live not disabled a little bit early and send messages to the instance. The fundamental issue is that HelixAdmin tool has not synchronization with Helix controller when changing messages of a participant.
   
   
   


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@helix.apache.org
For additional commands, e-mail: reviews-help@helix.apache.org