HDFS RAID actually uses map-reduce to asynchronously generate parity bits of
existing HDFS files.

thanks,
dhruba


On Fri, Jan 22, 2010 at 10:55 AM, Allen Wittenauer <[email protected]
> wrote:

> On 1/22/10 10:49 AM, "Dhruba Borthakur" <[email protected]> wrote:
> > I am moving the HDFS raid package to the mapreduce repository, via
> > MAPREDUCE-1388. I am going to be doing a "svn move" of these files so
> that
> > we do not lose the history of these files. Then I will apply the patch to
> > fix the compilation issues in the mapreduce repository (this means that
> the
> > mapreduce tree might not compile cleanly for a short duration of time).
>
>
> Am I reading this correctly:  we're moving HDFS RAID to MR because the
> *test
> code* fails?
>
>
>


-- 
Connect to me at http://www.facebook.com/dhruba

Reply via email to