[
https://issues.apache.org/jira/browse/FLINK-26372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17560983#comment-17560983
]
Feifan Wang commented on FLINK-26372:
-------------------------------------
Thanks [~roman] for explain patiently. I confused that we alread have method
'env.enableChangelogStateBackend(boolean)' which can set 'changelog.enabled'
from application code, I tested it and it works fine.
I guess I didn't understand the meaning of 'just reading "changelog.enabled"
from job configuration instead of cluster'. Can you explain where the 'job
configuration' come from ? I noticed
jobInformation in TaskExecutor.submitTask() has a method
'jobInformation.getJobConfiguration()', but I can't find where does the
returned configuration come from.
> Allow to configure Changelog Storage per program
> ------------------------------------------------
>
> Key: FLINK-26372
> URL: https://issues.apache.org/jira/browse/FLINK-26372
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Configuration, Runtime / State Backends
> Affects Versions: 1.15.0
> Reporter: Roman Khachatryan
> Assignee: Feifan Wang
> Priority: Major
> Fix For: 1.16.0
>
>
> It's currently possible to override state.backend.changelog.enabled per job,
> but it's not possible to override Changelog storage settings (i.e. writer
> settings).
> There should be 1) an API and 2) runtime support for that.
> See this
> [discussion|https://github.com/apache/flink/pull/16341#discussion_r663749681]
> and the corresponding
> [TODO|https://github.com/apache/flink/pull/16341/files#diff-2c21555dcab689ec27c0ab981852a2bfa787695fb2fe04b24c22b89c63d98b73R680].
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
