[
https://issues.apache.org/jira/browse/FLINK-14054?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16929459#comment-16929459
]
Jun Qin commented on FLINK-14054:
---------------------------------
I think It makes sense to have a cluster level default settings. Each job can
then overrides them via job level configurations.
In addition, I think enabling/disabling checkpointing on the fly is also a
useful feature. But this is probably worth another JIRA.
> Enable checkpointing via job configuration
> ------------------------------------------
>
> Key: FLINK-14054
> URL: https://issues.apache.org/jira/browse/FLINK-14054
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Checkpointing, Runtime / Configuration
> Reporter: Jun Qin
> Priority: Major
>
> Currently enabling checkpointing can only be done via the job code, see the
> following quote from this Flink
> [checkpointing|https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/stream/state/checkpointing.html#enabling-and-configuring-checkpointing]
> doc:
> {quote}By default, checkpointing is disabled. To enable checkpointing, call
> {{enableCheckpointing(n)}} on the {{StreamExecutionEnvironment}}, where _n_
> is the checkpoint interval in milliseconds.
> {quote}
> This makes enabling checkingpointing after the job code has been released
> difficult: one has to change and rebuild the job code.
> In addition, not only for developer, making checkpointing enabling
> configurable is also of interest for operation teams:
> * They may want to enable checkpointing for production but disable in test
> (e.g., to save storage space)
> * They may want to try out with and without checkpointing to evaluate the
> impact to the job behaviour and performance.
> Therefore, this request. Thanks.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
