If you set the number of reduce tasks to zero, the outputs of the mappers will be sent directly to the OutputFormat. You can debug your map phase of a job by disabling reduce and inspecting the mapper outputs, and then re-enable the reducer after you've got the mapping part of the job running correctly.
-- Aaron On Fri, Jul 2, 2010 at 10:10 AM, Ted Yu <[email protected]> wrote: > Minor correction - port is 50075: > http://10.32.56.159:50075/logs/userlogs/ > > On Fri, Jul 2, 2010 at 12:32 AM, Alberto Luengo Cabanillas < > [email protected]> wrote: > > > Hi! Supposing that you´re using a virtualized environment, you should be > > able to get to JobTracker Web Admin Interface from http:// > > <your_virtualized_ > > machine_ip>:50070/logs/userlogs. In case you´re running Hadoop from > inside > > a > > cluster frontend, use this machine´s IP. Inside this directory, it will > > exist another one called "userlogs" where you will see all map and reduce > > tasks launch by the JobTracker and the TaskTrackers. These tasks will > look > > like 'job_[m,r]_0_xxxxxx' (or similar). Inside of every one of them there > > will be three subdirectories (stdout, stdlog and stderr) which will > contain > > all your debugging messages. > > Inside my MapReduce development programs I use the Java *LOG* class in a > > sentence like this: > > > > private static final Log LOG = > > LogFactory.getLog(<your_class_name>.class.getName()); > > LOG.info("***********INSIDE MY REDUCER SETUP**********"); > > > > Besides these messages, a normal "System.out.println(...)" will also > appear > > in the above directories. > > > > Hope it helps! > > > > 2010/7/2 abc xyz <[email protected]> > > > > > Hi folks, > > > > > > I want to just display the output of my mapper on console or want to > log > > it > > > in > > > some file. System.out.println and Log4j's logger are not displaying the > > > output. > > > How can I see the intermediate results? > > > > > > Cheers > > > > > > > > > > > > > > > > > > > > > > > -- > > Alberto > > >
