What is JobTracker and what it does?

JobTracker is  master node in Map Reduce part of Hadoop architecture.Client submits the job to the JobTracker and then Map Reduce task is performed for calculating the actual result. MapReduce processing in Hadoop  is handled by the JobTracker and TaskTracker daemons. JobTracker places the jobs in a queue. A scheduler program  in the JobTracker picks up the jobs from it. The number of map tasks depends on the number of input splits on the data file. The number of reduce tasks depends on the value defined in the, setNumReduceTasks() method of the job object.JobTracker can not submit the result to the client until all the map and reduce task is completed.                                                                                                                           JobTracker  maintains the work of the TaskTrackers(Slave Nodes). It decides which TaskTracker will perform the map task based on its proximity to the data and which will perform reduce task. TaskTracker send heartbeat signal in every few second to JobTracker for informing that they are alive. If JobTracker don’t receive any heartbeat from any TaskTracker, it is considered as dead .

Note: The JobTracker runs on NameNode