jiajunwang commented on a change in pull request #1360:
URL: https://github.com/apache/helix/pull/1360#discussion_r490714047
##########
File path:
helix-core/src/test/java/org/apache/helix/integration/TestAlertingRebalancerFailure.java
##########
@@ -161,7 +170,8 @@ public void testParticipantUnavailable() throws Exception {
@Test(dependsOnMethods = "testParticipantUnavailable")
public void testTagSetIncorrect() throws Exception {
_gSetupTool.addResourceToCluster(CLUSTER_NAME, testDb, 5,
- BuiltInStateModelDefinitions.MasterSlave.name(),
RebalanceMode.FULL_AUTO.name());
Review comment:
It works in production. But it does not work in the test.
The reason is that our verifier does not read the controller cache (which is
only used by the DelayedAutoRebalancer).
So the controller only does one calculation, and later pipeline will only
read from the cache. But the verifier is calculating from scratch every time.
And since the AutoRebalanceStrategy is not stable, the verify result will never
converge.
If we change to CrushEd, then even the verifier computing from scratch, the
result would be the same. And our tests can pass.
----------------------------------------------------------------
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:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
