[
https://issues.apache.org/jira/browse/FLINK-14373?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gary Yao closed FLINK-14373.
----------------------------
Resolution: Fixed
1.10: e933a5cbc263be5041bd3723cd2400c871739733
> Enable ZooKeeperHighAvailabilityITCase to pass with scheduler NG
> ----------------------------------------------------------------
>
> Key: FLINK-14373
> URL: https://issues.apache.org/jira/browse/FLINK-14373
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination, Tests
> Affects Versions: 1.10.0
> Reporter: Zhu Zhu
> Assignee: Zhu Zhu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.10.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> ZooKeeperHighAvailabilityITCase currently fails with scheduler NG.
> There are 3 reasons for the failure:
> 1. it will invoke ExecutionGraph#failGlobal for global recovery but it is not
> ready for use in scheduler NG
> 2. the test relies on NumberOfFullRestartsGauge for restart count, which is
> not correct with fine grained recovery. This issue can be fixed by changing
> the failover metric to numberOfRestarts introduced in FLINK-14164.
> We need to support failGlobal in scheduler NG to make this case pass with
> scheduler NG. And then annotate it with AlsoRunWithSchedulerNG.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
