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-1946) du should be not called on every heartbeat
Date Fri, 28 Sep 2007 19:54:50 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-1946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12531134
] 

Konstantin Shvachko commented on HADOOP-1946:
---------------------------------------------

1. You need to initialize Command.lastTime in the constructor the way it was done in DF.
The idea is that the interval can be set to MAX_LONG, which means that I want to execute the
command once and never there after.
2. Could you please make a test case for that too.
3. I'd rename Command to ShellCommand. We have many other commands like UpgradeCommand, BlockCommand,
etc.
So it is confusing.

> du should be not called on every heartbeat
> ------------------------------------------
>
>                 Key: HADOOP-1946
>                 URL: https://issues.apache.org/jira/browse/HADOOP-1946
>             Project: Hadoop
>          Issue Type: Bug
>          Components: dfs
>    Affects Versions: 0.15.0
>            Reporter: Konstantin Shvachko
>            Assignee: Hairong Kuang
>            Priority: Blocker
>             Fix For: 0.15.0
>
>         Attachments: du.patch, du.patch
>
>
> Data-nodes run very slow. The reason is that du is executed on each and every heartbeat.
> Looks like the condition in DU.getUsed(), which triggers the refresh of cached values,
should be reversed.

-- 
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