You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "vinoyang (Jira)" <ji...@apache.org> on 2019/09/11 02:34:00 UTC

[jira] [Updated] (FLINK-12362) Remove container number config option for Flink on yarn job cluster mode

     [ https://issues.apache.org/jira/browse/FLINK-12362?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

vinoyang updated FLINK-12362:
-----------------------------
    Description: The {{-n}} config option for Flink on YARN has been deprecated and removed from documentation. However, there are still some legacy code exists in the codebase. We need to clean up them.  (was: Recently, the {{-n}} config option has been removed from the document, the commit is here: [https://github.com/apache/flink/commit/60507b8baf5fba999b7c2c45c150914f38f5ce38#diff-8ae07fac4c1f4137fc1ef062aef70115L100]

I can understand that maybe this option is useless for Flink session. But for job cluster mode, remove this option would confuse the users:
 * the {{-n}} has been deprecated and replaced with {{-p}}?
 * {{-p}} means the job level parallelism, but if users want to set a single operator's parallelism? how to calculate the number of containers
 * the {{-n}} option still exists in {{FlinkYarnSessionCli}}
 * we have customized the source code which broken the rule: one container one slot, so the {{-n}} is still useful

 

 )

> Remove container number config option for Flink on yarn job cluster mode
> ------------------------------------------------------------------------
>
>                 Key: FLINK-12362
>                 URL: https://issues.apache.org/jira/browse/FLINK-12362
>             Project: Flink
>          Issue Type: Improvement
>          Components: Deployment / YARN
>            Reporter: vinoyang
>            Assignee: vinoyang
>            Priority: Major
>
> The {{-n}} config option for Flink on YARN has been deprecated and removed from documentation. However, there are still some legacy code exists in the codebase. We need to clean up them.



--
This message was sent by Atlassian Jira
(v8.3.2#803003)