I have to admit that I don't know the official answer. The hack below seems working: - compile all 3 sub-projects; - copy everything in hdfs/build and mapreduce/build to common/build; - then run hadoop by the scripts in common/bin as before.
Any better idea? Nicholas Sze
