hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tao Jie (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4855) Should check if node exists when replace nodelabels
Date Mon, 05 Sep 2016 02:06:20 GMT

    [ https://issues.apache.org/jira/browse/YARN-4855?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15463872#comment-15463872

Tao Jie commented on YARN-4855:

Thank you for your comments, [~Naganarasimha].
acceptNode => isRMRegisteredNode or isValidNode, former seems to be better
The logic of verifying nodes is to compare *set* of requested nodes to  *set* of registered
nodes, and check if former set is subset of the latter. The method {{acceptNode}} here is
to compare *one* node to *another*(check if they are the same one). {{isRMRegiteredNode}}
seems like compare *one* node to a *set* of nodes. I will trying to improve like this(maybe
it is easier to understand).
seems like RMAdminCLI doesnt require both setYarnClient and createYarnClient.
{{setYarnClient}} is used only in testcase, where we need to have a mock yarnClient(like the
property {{localNodeLabelsManager}}). I tried to set {{yarnClient}} to protected, and assigned
it in test, it triggered another findbugs warning. I will trying to refine here, if you know
some best practices, please tell me.

> Should check if node exists when replace nodelabels
> ---------------------------------------------------
>                 Key: YARN-4855
>                 URL: https://issues.apache.org/jira/browse/YARN-4855
>             Project: Hadoop YARN
>          Issue Type: Improvement
>    Affects Versions: 2.6.0
>            Reporter: Tao Jie
>            Assignee: Tao Jie
>            Priority: Minor
>         Attachments: YARN-4855.001.patch, YARN-4855.002.patch, YARN-4855.003.patch, YARN-4855.004.patch,
YARN-4855.005.patch, YARN-4855.006.patch
> Today when we add nodelabels to nodes, it would succeed even if nodes are not existing
NodeManger in cluster without any message.
> It could be like this:
> When we use *yarn rmadmin -replaceLabelsOnNode --fail-on-unkown-nodes "node1=label1"*
, it would be denied if node is unknown.

This message was sent by Atlassian JIRA

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

View raw message