Hey, This has come up in past, please read: http://search-hadoop.com/m/KmgQh2WqnOB
On Tue, May 1, 2012 at 12:32 AM, Jianhui Zhang <jhzhang.em...@gmail.com> wrote: > Hi folks, > > Version: Hadoop 0.20.205. > > My reducer can be optimized if I can get a good estimate on how many > records are produced by the mappers, that is, if I can get the > MAP_OUTPUT_RECORDS counter (or its equivalent) in my reducer. However, > I tried and always got 0, guess the counters are not passed to the > reducer context. > > Anybody has encountered the same requirement? And any pointers on that? > > Appreciate your help. > > Thanks, > James -- Harsh J
