mapred-site.xml

MapReduce is a type of application that can run on the Hadoop 2.x framework. MapReduce configuration options are stored in the /opt/mapr/hadoop/hadoop-2.x.x/etc/hadoop/mapred-site.xml file and are editable by the root user. This file contains configuration information that overrides the default values for MapReduce parameters. Overrides of the default values for core configuration properties are stored in the MapR Data Platform Parameters file.

To override a default value for a property, specify the new value within the <configuration> tags, using the following format:

<property>
 <name> </name>
 <value> </value>
 <description> </description>
</property>

Configurations for MapReduce Applications

The configuration comprises the following parameters:

mapreduce.framework.name: Value: yarn; Description: Execution framework set to Hadoop YARN.
mapreduce.input.fileinputformat.split.maxblocknum: Value: 0; Description: Number of blocks that can be added to one split. A value of 0 means that a single split is generated per node.
This functionality requires a patch. To install patches, see Applying a Patch.
mapreduce.map.memory.mb: Value: 1024; Description: Larger resource limit for maps.

mapreduce.map.java.opts: Value: ‑Xmx1024M; Description: Larger heap-size for child jvms of maps.
mapreduce.reduce.memory.mb: Value: 3072; Description: Larger resource limit for reduces.

mapreduce.reduce.java.opts: Value: -Xmx2560m; Description: Larger heap-size for child jvms of reduces.

mapreduce.task.io.sort.mb: Value: 512; Description: Higher memory limit while sorting data for efficiency.
mapreduce.task.io.sort.factor: Value: 100; Description: More streams merged at once while sorting files.

mapreduce.reduce.shuffle.parallelcopies: Value: 50; Description: Higher number of parallel copies run by reduces to fetch outputs from very large number of maps.

Configurations for MapReduce JobHistory Server

The configuration comprises the following parameters:

mapr.localspill.expiration.date: Value: days; Description: Property to determine spill files expiration date in days. Default value is 30 days.

mapreduce.jobhistory.address: Value: MapReduce JobHistory Server host:port; Description: Default port is 10020.
mapreduce.jobhistory.webapp.address: Value: MapReduce JobHistory Server Web UI host:port; Description: Default port is 19888.
mapreduce.jobhistory.intermediate-done-dir: Value: /mr-history/tmp; Description: Directory where history files are written by MapReduce applications.
mapreduce.jobhistory.intermediate-done-scan-timeout: Value: milliseconds; Description: Timeout in milliseconds for rescanning the done_intermediate user directory to reduce JobHistory Server loading. Information about a job is received with a delay equal to the timeout. Adjust the setting based on the cluster load. Start with 5000 ms and increase timeout as needed.
NOTE This functionality requires a patch. To install patches, see Applying a Patch.
mapreduce.jobhistory.done-dir: Value: /mr-history/done; Description: Directory where history files are managed by the MapReduce JobHistory Server.

mapreduce.jobhistory.webapp.https.address: Value: Secure MapReduce JobHistory Server Web UI host:port (HTTPS); Description: Default port is 19890.

Sample Hadoop 2.x mapred-site.xml File

The following mapred-site.xml file defines values for two job history parameters.

<configuration>
  <property>
    <name>mapreduce.jobhistory.address</name>
    <value>__HS_IP__:10020</value>
  </property>
  <property>
    <name>mapreduce.jobhistory.webapp.address</name>
    <value>__HS_IP__:19888</value>
  </property>
</configuration>

Configuration for Apache Shuffle

You can disable Direct Shuffle and enable Apache Shuffle for MapReduce applications through the following settings:

mapreduce.job.shuffle.provider.services: Value: mapreduce_shuffle
mapreduce.job.reduce.shuffle.consumer.plugin.class: Value: org.apache.hadoop.mapreduce.task.reduce.Shuffle
mapreduce.job.map.output.collector.class: Value: org.apache.hadoop.mapred.MapTask$MapOutputBuffer
mapred.ifile.outputstream: Value: org.apache.hadoop.mapred.IFileOutputStream
mapred.ifile.inputstream: Value: org.apache.hadoop.mapred.IFileInputStream
mapred.local.mapoutput: Value: true
mapreduce.task.local.output.class: Value: org.apache.hadoop.mapred.YarnOutputFiles