What is the technique to know the Default scheduler in hadoop

Question

I have installed a multi node setup in 3 Ubuntu systems 12.04. I am using Hadoop1.2.1 in all three.Now i want to which scheduler is running by default??? How to check the default scheduler running in Hadoop1.2.1?

Answer 1

Default scheduler in hadoop is JobQueueTaskScheduler, which is a FIFO scheduler. As a default scheduler you need to refer the property mapred.jobtracker.taskScheduler in mapred-default.xml. If you want you can change the default scheduler to either CapacityScheduler or FairScheduler based on your requirement.

mapred-site.xml is used to override the default values inside mapred-default.xml, which can be found inside the configuration directory. You may not find mapred-default file in the configuration directory along with hadoop binary distribution(rpm,deb etc.), instead mapred-default.xml can be found directly inside the jar file hadoop-core-1.2.1.jar.

hackzon:~/hadoop-1.2.1$ jar -tvf hadoop-core-1.2.1.jar  | grep mapred-default.xml
 47324 Mon Jul 22 15:12:48 IST 2013 mapred-default.xml

These file is used in the below mentioned hadoop source files as an argument to addDefaultResource() method as

addDefaultResource("mapred-default.xml");   // First
addDefaultResource("mapred-site.xml");   // Second

Initially mapred-default.xml would be loaded, then mapred-site.xml. So that properties which need to be overridden can be specified inside mapred-site.xml

org.apache.hadoop.conf.Configuration.java
org.apache.hadoop.mapred.JobConf.java
org.apache.hadoop.mapred.TaskTracker.java
org.apache.hadoop.mapred.JobClient.java
org.apache.hadoop.mapred.JobTracker.java
org.apache.hadoop.mapred.JobHistoryServer.java