how to sort word count by value in hadoop? [duplicate]
You need to have a second mapreduce job. Unless you conclude on the the totals counts (which the first MR job does) how can you think of sorting by value (the counts of the words)? Logically not possible.