I'm running CDH4 MR1 and looking at performance metrics in preparation for migrating to a new hadoop environment.
Total time spent by all maps in occupied slots (ms) 0 0 3,801,573,232
Total time spent by all reduces in occupied slots (ms) 0 0 30,152,549
CPU time spent (ms) 0 0 5,151,794,220
Shouldn't CPU time be smaller than map millis+reduce millis because the map/reduce millis includes time when the CPU is both running and idle?