You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "Biao Liu (Jira)" <ji...@apache.org> on 2019/08/30 06:21:00 UTC

[jira] [Created] (FLINK-13904) Avoid competition between different rounds of checkpoint triggering

Biao Liu created FLINK-13904:
--------------------------------

             Summary: Avoid competition between different rounds of checkpoint triggering
                 Key: FLINK-13904
                 URL: https://issues.apache.org/jira/browse/FLINK-13904
             Project: Flink
          Issue Type: Sub-task
          Components: Runtime / Coordination
            Reporter: Biao Liu
             Fix For: 1.10.0


As a part of {{CheckpointCoordinator}} refactoring, I'd like to simplify the concurrent triggering logic.
The different rounds of checkpoint triggering would be processed sequentially. The final target is getting rid of timer thread and {{triggerLock}}.

Note that we can't avoid all competitions of triggering for now. There is still a competition between normal checkpoint triggering and savepoint triggering. We could avoid this competition by executing triggering in main thread. But it could not be achieved until all blocking operations are handled well in IO threads.



--
This message was sent by Atlassian Jira
(v8.3.2#803003)