hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vinayakumar B (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-8670) Better to exclude decommissioned nodes for namenode NodeUsage JMX
Date Tue, 21 Jul 2015 10:35:04 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8670?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14634924#comment-14634924
] 

Vinayakumar B commented on HDFS-8670:
-------------------------------------

Patch looks overall good.

have some nits in test, 
1. use try-finally after starting cluster and shutdown cluster in finally block.
2. Since DECOMMISSIONED and DECOMMISSION_INPROGRESS are enums, they are static, '==' also
can be used for comparison instead of {{equals(..)}}.


> Better to exclude decommissioned nodes for namenode NodeUsage JMX
> -----------------------------------------------------------------
>
>                 Key: HDFS-8670
>                 URL: https://issues.apache.org/jira/browse/HDFS-8670
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: Ming Ma
>            Assignee: J.Andreina
>         Attachments: HDFS-8670.1.patch, HDFS-8670.2.patch, HDFS-8670.3.patch, HDFS-8670.4.patch
>
>
> The namenode NodeUsage JMX has Max, Median, Min and Standard Deviation of DataNodes usage,
it currently includes decommissioned nodes for the calculation. However, given balancer doesn't
work on decommissioned nodes and sometimes we could have nodes stay in decommissioned states
for a long time; it might be better to exclude decommissioned nodes for the metrics calculation.



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

Mime
View raw message