You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@hudi.apache.org by "kwang (Jira)" <ji...@apache.org> on 2023/11/06 01:31:00 UTC

[jira] [Updated] (HUDI-6990) Configurable clustering task parallelism

     [ https://issues.apache.org/jira/browse/HUDI-6990?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

kwang updated HUDI-6990:
------------------------
    Summary: Configurable clustering task parallelism  (was: Spark clustering job reads records support control the parallelism)

> Configurable clustering task parallelism
> ----------------------------------------
>
>                 Key: HUDI-6990
>                 URL: https://issues.apache.org/jira/browse/HUDI-6990
>             Project: Apache Hudi
>          Issue Type: Improvement
>          Components: clustering
>            Reporter: kwang
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.0.0, 0.14.1
>
>         Attachments: after-subtasks.png, before-subtasks.png
>
>
> Spark executes clustering job will read clustering plan which contains multiple groups. Each group process many base files or log files. When we config param `
> hoodie.clustering.plan.strategy.sort.columns`, we read those files through spark's parallelize method, every file read will generate one sub task. It's unreasonable.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)