Adding to Bejoy's response, if you want to just keep the temp output of failed tasks, you can also use keep.failed.task.files.
Also, these config variable names are as per Hadoop 1.0. In trunk, they are called mapreduce.task.files.preserve.filepattern and mapreduce.task.files.preserve.failedtasks respectively. Thanks Hemanth On Thu, Sep 27, 2012 at 11:18 PM, Bejoy Ks <bejoy.had...@gmail.com> wrote: > Hi > > The temporary output from tasks can be preserved using the following > property'keep.task.files.pattern' > > http://books.google.co.in/books?id=drbI_aro20oC&pg=PA178&lpg=PA178&dq=keep.task.files.pattern&source=bl&ots=tZAmxgm_j4&sig=Guc0bh2BQzlbMqOADtic5WciIz0&hl=en&sa=X&ei=zI9kULbDM8zhrAe3jYH4BA&ved=0CFoQ6AEwBg#v=onepage&q=keep.task.files.pattern&f=false > > > On Thu, Sep 27, 2012 at 10:49 PM, Pedro Sá da Costa <psdc1...@gmail.com> > wrote: >> >> Hi, >> >> Is there a parameter to avoid to clean the temporary data created during >> the job execution after the job end? >> >> -- >> Best regards, >> >
