[ 
https://issues.apache.org/jira/browse/HADOOP-3479?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12604916#action_12604916
 ] 

Steve Loughran commented on HADOOP-3479:
----------------------------------------

'consistent': how the things are configured *and* how the configs are read are 
similar. You don't want to go the java webapp way with an ear.xml, a war.xml, a 
persistence xml file, each with their own syntax and a need to try and keep the 
things coherent.

As of 5 minutes ago, I have checked in the outline slides for the august UK 
talk, in which I talk about managing the config and lifecycle of Hadoop nodes:

http://smartfrog.svn.sourceforge.net/viewvc/*checkout*/smartfrog/trunk/core/components/hadoop/doc/managing_hadoop.ppt

To handle hadoop configuration we had to subclass JobConf (as too much does 
instanceof JobConf tests), binding to our own configuration source *and* 
replicating any private methods of Configuration needed to serialize the 
configuration into the wire format, which is an XML document:

http://smartfrog.svn.sourceforge.net/viewvc/smartfrog/trunk/core/components/hadoop/src/org/smartfrog/services/hadoop/conf/ManagedConfiguration.java?view=markup

There was some discussion on HADOOP-24 on doing things to Configuration; what 
would seem best would be to let different things be a source of configuration 
data, with the actual Configuration class or subclass (including JobConf and 
HBaseConf) getting their input from that source. 

> Implement configuration items useful for Hadoop resource manager (v1)
> ---------------------------------------------------------------------
>
>                 Key: HADOOP-3479
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3479
>             Project: Hadoop Core
>          Issue Type: New Feature
>            Reporter: Hemanth Yamijala
>            Assignee: Hemanth Yamijala
>         Attachments: 3479.patch
>
>
> HADOOP-3421 lists requirements for a new resource manager for Hadoop. 
> Implementation for these will require support for new configuration items in 
> Hadoop. This JIRA is to define such configuration, and track it's 
> implementation.

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

Reply via email to