This leads to a race condition in which the mapReduce command can attempt to access the Collection's cursor manager after it has been destroyed. This can lead to the server shutting down with a segmentation fault, though I suspect the probability of such an event occurring is low.
- is duplicated by
- 
                    SERVER-23621 mapReduce with output collection writes incorrect 'ns' to profiler -         
- Closed
 
-         
- is related to
- 
                    SERVER-27362 MapReduce can attempt to deregister a cursor without holding a collection lock -         
- Closed
 
-