Shi, This states "Of course, the framework discards the sub-directory of unsuccessful task-attempts." http://hadoop-karma.blogspot.com/2011/01/hadoop-cookbook-how-to-write.ht ml So yes, the missing directory is likely a failure. If you can, narrow the problem down by looking at sections of your data on a smaller number of mappers. Look at just the tail of your input and a couple of mappers and then compare that to your program run without hadoop. Kevin -----Original Message----- From: Shi Yu -- Kevin Leach
↧