[
https://issues.apache.org/jira/browse/YARN-4434?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15048225#comment-15048225
]
ASF GitHub Bot commented on YARN-4434:
--------------------------------------
Github user aajisaka commented on the pull request:
https://github.com/apache/hadoop/pull/62#issuecomment-163142562
Thank you for the pull request. I reviewed the patch (A) and the another
patch in YARN-4434 jira (B) and decided to commit the patch (B) because the
patch (B) replaces "i.e. the entire disk" with "i.e. 90% of the disk" as well.
> NodeManager Disk Checker parameter documentation is not correct
> ---------------------------------------------------------------
>
> Key: YARN-4434
> URL: https://issues.apache.org/jira/browse/YARN-4434
> Project: Hadoop YARN
> Issue Type: Bug
> Components: documentation, nodemanager
> Affects Versions: 2.6.0, 2.7.1
> Reporter: Takashi Ohnishi
> Assignee: Weiwei Yang
> Priority: Minor
> Fix For: 2.8.0, 2.6.3, 2.7.3
>
> Attachments: YARN-4434.001.patch, YARN-4434.branch-2.6.patch
>
>
> In the description of
> yarn.nodemanager.disk-health-checker.max-disk-utilization-per-disk-percentage,
> it says
> {noformat}
> The default value is 100 i.e. the entire disk can be used.
> {noformat}
> But, in yarn-default.xml and source code, the default value is 90.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
