TisonKun commented on issue #4376: [FLINK-6521] Add per job cleanup methods to HighAvailabilityServices URL: https://github.com/apache/flink/pull/4376#issuecomment-447238896 @tillrohrmann @zjureel I think the functionality is implemented occasionally by #6587 [FLINK-10011](https://issues.apache.org/jira/browse/FLINK-10011) However, it is still a valid question that who is the proper actor to do the clean-up job. As for `SubmittedJobGraph`, it is managed by `Dispatcher`, but the `RunningJobsRegistry` is tricky that both `JobManagerRunner` and `Dispatcher` can write it. Under the topic "per job clean up", I'd like to raise a question that how flink considered the status of a job? If we said that "per job clean up" is `Dispatcher`'s responsibility, then we should prevent JM from writing such a `RunningJobsRegistry` and also it means that the status on `Dispatcher` is what we(users) think that of a certain job.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
