hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Konstantin Shvachko (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-4701) Improve datanode decommission monitoring performance
Date Sat, 22 Nov 2008 00:42:44 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-4701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12649869#action_12649869
] 

Konstantin Shvachko commented on HADOOP-4701:
---------------------------------------------

Raghu, my idea was that currently {{NumberReplicas}} counts live and decommissioned replicas
disjointly, and there is juggling with these two numbers all over the replication code. You
are saying there will be no decommissioned replicas and therefore all this code should be
revised. This is a big change to the very core of the name-node code and this is what scares
me even more than the change in the behavior.

> Improve datanode decommission monitoring performance
> ----------------------------------------------------
>
>                 Key: HADOOP-4701
>                 URL: https://issues.apache.org/jira/browse/HADOOP-4701
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: dfs
>            Reporter: Tsz Wo (Nicholas), SZE
>
> The datanode decommission monitoring is not efficient.  There are rooms for improvement.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message