how to profile hadoop tasks with yourkit

Question

I'm trying to profile the memory usage of my hadoop job.

Could someone provide a step by step how-to on how to monitor hadoop tasks with yourkit - including setup?

Answer 1

All you have to do is add the following entry to your mapred-site.xml file(which if found in $HADOOP_HOME/conf/, where $HADOOP_HOME is your Hadoop installation directory):

<property>
  <name>mapred.child.java.opts</name>
  <value>
 -agentpath:{yourkit installation directory}/bin/linux-x86-64/libyjpagent.so=tracing,dir={output directory}
 </value>
</property>

If you are running on a platform different from linux-x86-64, you might need to change above value to match your platform(see this for details)

You can pass any of the options listed here to the profiler agent

This will create a number of snapshots, one for each Child process in the specified output directory