[
https://issues.apache.org/jira/browse/FLINK-39937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18089655#comment-18089655
]
Robert Metzger commented on FLINK-39937:
----------------------------------------
This is currently not supported directly via the Flink Kubernetes operator.
However, you can build a solution for this yourself using the MiniCluster class
inside Flink. It will launch all required Flink components in a single JVM.
> Merge jobmanager and taskmanager into the same JVM
> --------------------------------------------------
>
> Key: FLINK-39937
> URL: https://issues.apache.org/jira/browse/FLINK-39937
> Project: Flink
> Issue Type: Improvement
> Components: API / Core
> Affects Versions: 2.4.0
> Reporter: melin
> Priority: Major
>
> When using flink to synchronize data, such as reading data from MySQL and
> writing it into a Hive table, two resources (pods) are required: the
> jobmanager and the taskmanager. If the taskmanager value is 1, it is possible
> to combine the jobmanager and the taskmanager within the same JVM, requiring
> only one pod, which can save half of the cost.
> [https://github.com/apache/spark/issues/56353]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
