hadoop-yarn-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From 牛兆捷 <nzjem...@gmail.com>
Subject Re: About container preemption
Date Mon, 24 Jun 2013 16:29:44 GMT
Thanks ~

2013/6/24 Harsh J <harsh@cloudera.com>

> Hi,
>
> You may track https://issues.apache.org/jira/browse/MAPREDUCE-5269 for
> progress on Reduce end checkpointing.
>
> On Sat, Jun 22, 2013 at 12:12 PM, 牛兆捷 <nzjemail@gmail.com> wrote:
> > Hi Chris:
> > I don't find the related checkpoint implementation for shuffle and reduce
> > phase of MapReduce.
> > Are the code open now?
> >
> >
> > 2013/6/22 Chris Douglas <cdouglas@apache.org>
> >
> >> The APIs for checkpointing are in branch-2, but support for the
> >> capacity scheduler and the MapReduce implementation are making their
> >> way through review.
> >>
> >> As of now, the checkpointing is for the shuffle and reduce phases,
> >> only. Map checkpointing requires cooperation with the RecordReader
> >> (i.e., the symmetric operation to create an InputSplit from a
> >> RecordReader) which has not been implemented. -C
> >>
> >> On Thu, Jun 20, 2013 at 10:29 PM, 牛兆捷 <nzjemail@gmail.com> wrote:
> >> > https://issues.apache.org/jira/browse/YARN-45 support work-preserve
> >> > preemption
> >> >
> >> > Has the AM support container's state checkpoint now?
> >> > Does the checkpoint support map, shuffle and reduce phase?
> >> >
> >> > --
> >> > *Sincerely,*
> >> > *Zhaojie*
> >> > *
> >> > *
> >>
> >
> >
> >
> > --
> > *Sincerely,*
> > *Zhaojie*
> > *
> > *
>
>
>
> --
> Harsh J
>



-- 
*Sincerely,*
*Zhaojie*
*
*

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message