You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hadoop.apache.org by Harsh J <ha...@cloudera.com> on 2013/09/04 08:08:52 UTC

Re: Hadoop Yarn

This is sorta a regression with YARN+MR2. In MR1 we've had the
administrative ability to lock down changes to mapred.child.java.opts
on a per node basis, as Tasks would load the TT config and they could
supply finalized arguments (controllable per-node when needed).

Given the app behavior of MR2, I am not sure if its still a good idea
to do, as we wouldn't expect all cluster nodes to even carry the MR
config anymore.

We could have the NM inspect+inject or override container opts though, thoughts?

On Fri, Aug 30, 2013 at 4:07 AM, Vinod Kumar Vavilapalli
<vi...@apache.org> wrote:
>
> You'll have to change the MapReduce code. What options are you exactly
> looking for and why should they be only applied on some nodes? Some kind of
> sampling?
>
> More details can help us help you.
>
> Thanks,
> +Vinod Kumar Vavilapalli
> Hortonworks Inc.
> http://hortonworks.com/
>
> On Aug 29, 2013, at 1:59 PM, Rajesh Jain wrote:
>
> I have some jvm options which i want to configure only for a few nodes in
> the cluster using Hadoop yarn. How do i di it. If i edit the mapred-site.xml
> it gets applied to all the task jvms. I just want handful of map jvms to
> have that option and other map jvm not have that options.
>
> Thanks
> Rajesh
>
> Sent from my iPhone
>
>
>
> CONFIDENTIALITY NOTICE
> NOTICE: This message is intended for the use of the individual or entity to
> which it is addressed and may contain information that is confidential,
> privileged and exempt from disclosure under applicable law. If the reader of
> this message is not the intended recipient, you are hereby notified that any
> printing, copying, dissemination, distribution, disclosure or forwarding of
> this communication is strictly prohibited. If you have received this
> communication in error, please contact the sender immediately and delete it
> from your system. Thank You.



-- 
Harsh J