Reputation: 2361
I have a C++ based MapReduce jobs and I'm using Hadoop streaming.
However the maximum number of reducers per node end up being 7
even if I set them to 18
in the command line configuration as mapred.tasktracker.reduce.tasks.maximum=18
;
Is anything else stopping the reducer from emitting more reduce tasks ?
Upvotes: 1
Views: 2111
Reputation: 30089
After amending the mapred.tasktracker.reduce.tasks.maximum
property, are you restarting the task trackers in your cluster? You should be able to go to the Job Tracker web ui page and confirm that each task tracker now has 18 reducer slots configured
Upvotes: 1