MR1 has the concept of "map and reduce slots" (i.e. different types of
slots) - but it does not run maps in reducer allocated slots - they
are exclusive for reducers of any job to run.
YARN, on which MR2 runs, has no notion of dedicated slots (for M or R types).
On Sat, Dec 29, 2012 at 6:51 PM, P
MapReduce framework has map and reduce slots, that are used to track which
tasks are running. When map tasks are just running, the reduce slots that
the job have will be filled by map tasks?
--
Best regards,
