You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Yu Li (Jira)" <ji...@apache.org> on 2020/05/21 10:23:00 UTC
[jira] [Closed] (FLINK-14489) Check and make sure state backend
fits into TaskExecutor memory configuration
[ https://issues.apache.org/jira/browse/FLINK-14489?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Yu Li closed FLINK-14489.
-------------------------
Resolution: Later
We had some discussion (in this [google doc|https://docs.google.com/document/d/12HL7qd3fF3nLOipOqXGkYk8cA-pIPYtDCuRiQEKMsX8/edit?usp=sharing]) but since the current implementation of Heap backend has no memory accounting logic, we actually only need to take care of the {{RocksDBStateBackend}}. We will get back and review this again when there're more types of backends which requires a more general design.
> Check and make sure state backend fits into TaskExecutor memory configuration
> -----------------------------------------------------------------------------
>
> Key: FLINK-14489
> URL: https://issues.apache.org/jira/browse/FLINK-14489
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / State Backends, Runtime / Task
> Reporter: Yu Li
> Assignee: Yu Li
> Priority: Major
>
> [FLIP-49|https://cwiki.apache.org/confluence/display/FLINK/FLIP-49%3A+Unified+Memory+Configuration+for+TaskExecutors] and [FLINK-13980|https://issues.apache.org/jira/browse/FLINK-13984] has proposed a unified memory configuration for {{TaskExecutor}}, we should find a proper way to suit state backend memory management into it. This sub-task will track the discussion and solution on this topic.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)