[
https://issues.apache.org/jira/browse/FLINK-14765?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16979378#comment-16979378
]
Andrey Zagrebin commented on FLINK-14765:
-----------------------------------------
STATE_UPDATER looks to be used only in main thread, so it could be just a
direct assignment to the volatile state.
The read path is more complicated. It looks like CheckpointCoordinator does not
fully operate in the main thread yet, so we have to keep volatile.
> Remove STATE_UPDATER in Execution
> ---------------------------------
>
> Key: FLINK-14765
> URL: https://issues.apache.org/jira/browse/FLINK-14765
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Reporter: vinoyang
> Priority: Major
>
> After making access to ExecutionGraph single-threaded in FLINK-11417, we can
> simplify execution state update and get rid of STATE_UPDATER while about the
> volatile at the moment which could be further investigated.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
