Reducer's Performance in Hadoop

[Or Raz]

Is there an easy way (API way) to measure the reducer's (reduce task) performance in Apache Hadoop (version 2.6.5 if it matters).

1. Measuring the completion time (from starting the task till it finishes) of an reduce task? without measuring the time before and after in JAVA?
   
2. Finding where the reduce tasks have been activated before or after they run? without (again) using any java code in the reducer class.
   

In addition, is there any (easy) way to insert more variables to "inside functions"? which we sometimes override, for instance getPartition inside Partitioner Class (or use an unset/unused variable that I can modify easily using API, for example, setNumReduceTasks)