Spark combineByKey different result between local and cluster

Question

I\'m trying to implement combineByKey function from a spark on java.

 JavaPairRDD, Integer> result = myPairRdd.com