[
https://issues.apache.org/jira/browse/FLINK-4947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16136704#comment-16136704
]
Gábor Hermann commented on FLINK-4947:
--------------------------------------
Would this make every configuration available in the CLI what's already
available in flink-conf.yaml? Similarly to Spark's {{spark-submit --conf
key=value}}? E.g. we could do {{flink run -m yarn-cluster ... --conf
containerized.heap-cutoff-ratio=0.4}}, and it would override the config found
in flink-conf.yaml.
Would it be hard to implement this? I could start working on this if you don't
see any serious caveats.
> Make all configuration possible via flink-conf.yaml and CLI.
> ------------------------------------------------------------
>
> Key: FLINK-4947
> URL: https://issues.apache.org/jira/browse/FLINK-4947
> Project: Flink
> Issue Type: Improvement
> Components: DataStream API
> Reporter: Jamie Grier
>
> I think it's important to make all configuration possible via the
> flink-conf.yaml and the command line.
> As an example: To enable "externalizedCheckpoints" you must actually call
> the StreamExecutionEnvironment#enableExternalizedCheckpoints() method from
> your Flink program.
> Another example of this would be configuring the RocksDB state backend.
> I think it important to make deployment flexible and easy to build tools
> around. For example, the infrastructure teams that make these configuration
> decisions and provide tools for deploying Flink apps, will be different from
> the teams deploying apps. The team writing apps should not have to set all
> of this lower level configuration up in their programs.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
