You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Roman Khachatryan (Jira)" <ji...@apache.org> on 2021/06/24 13:14:00 UTC
[jira] [Updated] (FLINK-23137) State ownership: disable discarding
and ref counting state of ChangelogStateBackendHandle on JM
[ https://issues.apache.org/jira/browse/FLINK-23137?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Roman Khachatryan updated FLINK-23137:
--------------------------------------
Summary: State ownership: disable discarding and ref counting state of ChangelogStateBackendHandle on JM (was: State ownership: disable discarding state of ChangelogStateBackendHandle on JM)
> State ownership: disable discarding and ref counting state of ChangelogStateBackendHandle on JM
> -----------------------------------------------------------------------------------------------
>
> Key: FLINK-23137
> URL: https://issues.apache.org/jira/browse/FLINK-23137
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / State Backends
> Reporter: Roman Khachatryan
> Priority: Major
> Fix For: 1.14.0
>
>
> TM should own changelog backend state to prevent re-uploading state on checkpoint abortion (or missing confirmation). A simpler solution to only own aborted state is less maintanable in the long run.
> For that, on JM side, ChangelogStateBackendHandle shouldn't discard and nor register and yhared state.
>
> Ticket for TM-side: FLINK-23139
>
> See [state ownership design doc|https://docs.google.com/document/d/1NJJQ30P27BmUvD7oa4FChvkYxMEgjRPTVdO1dHLl_9I/edit?usp=sharing]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)