According to this link http://wiki.apache.org/hadoop/HowToDebugMapReducePrograms
I am trying to find out where the downlink.data file is actually stored. I have set the variable keep.failed.task.files to true in mapred-site.xml, and hadoop.pipes.command-file.keep to true in the config.xml file that I pass to the pipes mapreduce task. Yet, there is no downlink.data file that I can locate either locally or on hdfs. I assume from the instructions for debugging that I should run the entire task in distributed mapreduce mode, collect the input in downlink.data and then run the mapreduce task with just the downlink.data. I am assuming 'task working directory' refers to where I ran 'hadoop pipes ...'. I've done a 'locate downlink.data' but nothing found. Am I making some incorrect assumption here? Or anything else needs to be set? I am running Hadoop 1.0.4 Thanks, Alice
