Let us discuss the few differences between Hadoop and MongoDB MapReduce.
Hadoop MapReduce is capable of using all available cores while MongoDB MapReduce is single threaded.
Hadoop MR cannot collocate with data while MongoDB MR can.
Hadoop MR has millions of engine/hours and can work with really massive data while MongoDB MR is slightly less expandable.
Hadoop MR can handle higher Frameworks like Pig, Sqoop and Hive etc while MongoDB cant.
Hadoop MR is Mainstream and has a lot of community support while MongoDB, I doubt.
Bottom line is that MongoDB MR can be used for Light and something simple.