I tried to use one or another for secondary sort -- both options work fine
-- I get combined sorted result in reduce() iterator

Also I noticed that if I set both of them at the same time
then KeyComparatorClass.compare(O1, O2) never called,  hadoop calls
only ValueGroupingComparator.compare()

I run my tests on single node instillation.

Please help me understand the diff btw these two comparators.