You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Sameer Paranjpye <sa...@yahoo-inc.com> on 2006/06/05 20:16:08 UTC
Re: Reduce Copy - waiting on task tracker
You might want to try out the 0.3 release. A lot of work has gone into
the copy implementation and you should be seeing both better performance
and more detail on the status page about the state of the copy for each
reduce.
Vijay Murthi wrote:
> Hi all,
>
> Many reduce tasks seems stall for a long time on "reduce > copy" status
> especially they all wait on one map task transfer example
> task_0048_m_000120_0@node 10:50040. So the status is like
> "reduce > copy > task_0048_m_000120_0@node 10:50040" for somewhat 25% of
> the nodes whereas rest have even completed.
>
> Could it because the tasktracker died on that node or it could be
> because tasktracker has nothing to transfer for these nodes and these
> 25% nodes keep waiting on forever?
>
> Thanks,
> VJ
>
>
>