[ 
https://issues.apache.org/jira/browse/HADOOP-13225?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15331733#comment-15331733
 ] 

Allen Wittenauer commented on HADOOP-13225:
-------------------------------------------

I was going through some old email today and ended up thinking about this JIRA 
this morning.  

A few months ago, someone asked me how easy it would be to run the Hadoop 
daemons in a cgroup with 3.x.  I mentioned that it would be trivial: just 
replace the java execution functions.

One of the key centerpieces of the shell script rewrite was the ability to 
replace functions.  This means that if an end user doesn't like how we do 
something, they can replace it with their own. 

This JIRA is a *great* example of that in action.  One user wants numactl and 
another wants cgexec.  We really can't support both without making a lot of "if 
this then this, otherwise, if this then this, and finally if this then this." 
type decisions in the shell code.

What we really should be doing here is providing this as an additional example 
in hadoop-user-functions.sh.example.  It's a really good one because it's 
trivial to write (at least for the non-secure case), useful for a sub-class of 
users, and a great template for other sub-classes of users to do their own 
logic (e.g., pfexec on Solaris).

With that said, I'll volunteer to write it up if another committer is actually 
willing to review it.

> Allow java to be started with numactl
> -------------------------------------
>
>                 Key: HADOOP-13225
>                 URL: https://issues.apache.org/jira/browse/HADOOP-13225
>             Project: Hadoop Common
>          Issue Type: New Feature
>          Components: scripts
>            Reporter: Dave Marion
>            Assignee: Dave Marion
>         Attachments: HDFS-10370-1.patch, HDFS-10370-2.patch, 
> HDFS-10370-3.patch, HDFS-10370-branch-2.004.patch, HDFS-10370.004.patch
>
>
> Allow numactl constraints to be applied to the datanode process. The 
> implementation I have in mind involves two environment variables (enable and 
> parameters) in the datanode startup process. Basically, if enabled and 
> numactl exists on the system, then start the java process using it. Provide a 
> default set of parameters, and allow the user to override the default. Wiring 
> this up for the non-jsvc use case seems straightforward. Not sure how this 
> can be supported using jsvc.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org

Reply via email to