1 - In JobTracker in the Hadoop Mapreduce 1.0.3, there's a new JobToken.
What's the purpose of the JobToken?
2 - I also notice that the input files has now some metafiles. It seems
that the way how tasks get the input files for the map tasks are completely
different from what hadoop 0.20.0 do. With the new version the input file
name isn't given directly to the map tasks. Can someone give me an insight
how the map tasks know which is the file name and path of the input split?