hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Billie Rinaldi (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-5883) Avoid or eliminate expensive YARN get all applications call
Date Tue, 22 Nov 2016 20:45:58 GMT

    [ https://issues.apache.org/jira/browse/YARN-5883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15687842#comment-15687842
] 

Billie Rinaldi commented on YARN-5883:
--------------------------------------

[~gsaha], the patch looks good overall. I would suggest not creating the version and description
tags when they are null. I am wondering about the usage of new HashSet<>(1) instead
of Collections.singleton(value). Do you have any feelings on that? If we're in the habit of
sizing our sets, maybe createApplicationTags should use new HashSet<>(3) as well.

> Avoid or eliminate expensive YARN get all applications call
> -----------------------------------------------------------
>
>                 Key: YARN-5883
>                 URL: https://issues.apache.org/jira/browse/YARN-5883
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Gour Saha
>            Assignee: Gour Saha
>             Fix For: yarn-native-services
>
>         Attachments: YARN-5883-yarn-native-services.001.patch
>
>
> YARN _getAllApplications()_ call with no param or with application types _getApplications(Set<String>
applicationTypes)_ are very costly in a cluster with hundreds or thousands of Slider apps
running.
> YARN now has the API to filter by application tags - YARN-4491. We should be able to
use this to avoid these expensive calls.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org


Mime
View raw message