Running Apache Spark on YARN with Docker

Running Apache Spark on YARN with Docker

Introduction

Apache Spark is a fast and general engine for large-scale data processing. Originally developed at the University of California, Berkeley's AMPLab, the Spark codebase was later donated to the Apache Software Foundation that has maintained it since. Spark provides an interface for programming entire clusters with implicit data parallelism and fault-tolerance.

Environment Versions

Ubuntu Server 14.04.4 LTS (Trusty Tahr) 64-bit
Docker 1.10.3
Hadoop 2.6.0
Spark 1.6.0

Install Docker

Update package information, ensure that APT works with the https method, and that CA certificates are installed.

$ apt-get update
$ apt-get install apt-transport-https ca-certificates

Add the new GPG key.

$ sudo apt-key adv --keyserver hkp://p80.pool.sks-keyservers.net:80 --recv-keys 58118E89F3A912897C070ADBF76221572C52609D

Open the /etc/apt/sources.list.d/docker.list file in your favorite editor.

If the file doesn't exist, create it.

Remove any existing entries.

Add an entry for your Ubuntu operating system.

deb https://apt.dockerproject.org/repo ubuntu-trusty main

Save and close the /etc/apt/sources.list.d/docker.list file.

Update the APT package index and install Docker:

$ sudo apt-get update && sudo apt-get install docker-engine

Running docker without sudo

$ sudo gpasswd -a ${USER} docker

Restart the docker daemon.

$ sudo service docker restart

Verify docker is installed correctly.

$ docker run hello-world

Pull Spark Images

Docker pull Spark image from Docker Hub:

$ docker pull sequenceiq/spark:1.6.0

Create Container

We will build a multi-node cluster in which one is master and the other is slaves1 and slave2, create three continers in different terminal.

Create master:

$ docker run --name master -it -p 8088:8088 -p 8042:8042 -p 8085:8080 -p 4040:4040 -p 7077:7077 -p 2022:22  -v /data:/data  -h master sequenceiq/spark:1.6.0 bash

Create slaves:

$ docker run --name slave1 -it -h slave1 sequenceiq/spark:1.6.0 bash
$ docker run --name slave2 -it -h slave2 sequenceiq/spark:1.6.0 bash

Hadoop Configuration

Stop the Hadoop multi-node cluster on all nodes:

# $HADOOP_COMMON_HOME/sbin/stop-dfs.sh && $HADOOP_COMMON_HOME/sbin/stop-yarn.sh

Update /etc/hosts on every nodes:

# vi /etc/hosts

Put the alias to the ip addresses of all the nodes:

172.17.0.2      master  # IP address of the master node
172.17.0.3      slave1  # IP address of the slave2 node
172.17.0.4      slave2  # IP address of the slave2 node

Configuration Hadoop slaves file on all nodes:

# vi $HADOOP_CONF_DIR/slaves
slave1
slave2

Create name, data and tmp folder for Hadoop on all nodes:

# mkdir -pv $HADOOP_COMMON_HOME/dfs/name
# mkdir -pv $HADOOP_COMMON_HOME/dfs/data
# mkdir -pv $HADOOP_COMMON_HOME/tmp

Configuration Hadoop core-site.xml file on all nodes:

# vi $HADOOP_CONF_DIR/core-site.xml

Change sandbox to master and add a hadoop.tmp.dir property:

<configuration>
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://master:9000</value>
    </property>
    <property>
        <name>io.file.buffer.size</name>
        <value>131702</value>
    </property>
    <property>
        <name>hadoop.tmp.dir</name>
        <value>/usr/local/hadoop/tmp</value>
    </property>
</configuration>

Configuration Hadoop hdfs-site.xml file on all nodes:

$ vi $HADOOP_CONF_DIR/hdfs-site.xml

Set dfs.replication value from 1 to 3:

<configuration>
    <property>
        <name>dfs.replication</name>
        <value>3</value>
    </property>
    <property>
        <name>dfs.namenode.name.dir</name>
        <value>/usr/local/hadoop/dfs/name</value>
    </property>
    <property>
        <name>dfs.datanode.data.dir</name>
        <value>/usr/local/hadoop/dfs/data</value>
    </property>
    <property>
        <name>dfs.namenode.secondary.http-address</name>
        <value>master:9001</value>
    </property>
    <property>
        <name>dfs.webhdfs.enabled</name>
        <value>true</value>
    </property>
</configuration>

Configuration Hadoop mapred-site.xml file on all nodes:

$ vi $HADOOP_CONF_DIR/mapred-site.xml

Change sandbox to master:

<configuration>
    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
    </property>
    <property>
        <name>mapreduce.jobhistory.address</name>
        <value>master:10020</value>
    </property>
    <property>
        <name>mapreduce.jobhistory.webapp.address</name>
        <value>master:19888</value>
    </property>
</configuration>

Configuration Hadoop yarn-site.xml file on all nodes:

$ vi $HADOOP_CONF_DIR/yarn-site.xml

Change sandbox to master:

<configuration>
    <property>
        <name>yarn.nodemanager.aux-services</name>
        <value>mapreduce_shuffle</value>
    </property>
    <property>
        <name>yarn.nodemanager.auxservices.mapreduce.shuffle.class</name>
        <value>org.apache.hadoop.mapred.ShuffleHandler</value>
    </property>
    <property>
        <name>yarn.resourcemanager.address</name>
        <value>master:8032</value>
    </property>
    <property>
        <name>yarn.resourcemanager.scheduler.address</name>
        <value>master:8030</value>
    </property>
    <property>
        <name>yarn.resourcemanager.resource-tracker.address</name>
        <value>master:8031</value>
    </property>
    <property>
        <name>yarn.resourcemanager.admin.address</name>
        <value>master:8033</value>
    </property>
    <property>
        <name>yarn.resourcemanager.webapp.address</name>
        <value>master:8088</value>
    </property>
    <property>
        <name>yarn.nodemanager.resource.memory-mb</name>
        <value>2048</value>
    </property>
    <property>
        <name>yarn.nodemanager.resource.cpu-vcores</name>
        <value>1</value>
    </property>
    <property>
        <name>yarn.application.classpath</name>
        <value>/usr/local/hadoop/etc/hadoop, /usr/local/hadoop/share/hadoop/common/*, /usr/local/hadoop/share/hadoop/common/lib/*, /usr/local/hadoop/share/hadoop/hdfs/*, /usr/local/hadoop/share/hadoop/hdfs/lib/*, /usr/local/hadoop/share/hadoop/mapreduce/*, /usr/local/hadoop/share/hadoop/mapreduce/lib/*, /usr/local/hadoop/share/hadoop/yarn/*, /usr/local/hadoop/share/hadoop/yarn/lib/*</value>
    </property>
    <property>
        <description>
          Number of seconds after an application finishes before the nodemanager's
          DeletionService will delete the application's localized file directory
          and log directory.

          To diagnose Yarn application problems, set this property's value large
          enough (for example, to 600 = 10 minutes) to permit examination of these
          directories. After changing the property's value, you must restart the
          nodemanager in order for it to have an effect.

          The roots of Yarn applications' work directories is configurable with
          the yarn.nodemanager.local-dirs property (see below), and the roots
          of the Yarn applications' log directories is configurable with the
          yarn.nodemanager.log-dirs property (see also below).
        </description>
        <name>yarn.nodemanager.delete.debug-delay-sec</name>
        <value>600</value>
    </property>
</configuration>

Format NameNode on master node:

# hdfs namenode -format

Spark Configuration

Configuration Spark slaves file on all nodes:

# cp $SPARK_HOME/conf/slaves.template $SPARK_HOME/conf/slaves && vi $SPARK_HOME/conf/slaves

Add following lines in file:

slave1
slave2

Configuration Spark spark-env.sh file on all node:

# cp $SPARK_HOME/conf/spark-env.sh.template $SPARK_HOME/conf/spark-env.sh && vi $SPARK_HOME/conf/spark-env.sh

Add following lines in file:

export JAVA_HOME=/usr/java/default
export SPARK_MASTER_IP=master
export SPARK_WORKER_CORES=1
export SPARK_WORKER_INSTANCES=1
export SPARK_MASTER_PORT=7077
export SPARK_WORKER_MEMORY=1g
export MASTER=spark://${SPARK_MASTER_IP}:${SPARK_MASTER_PORT}

Configuration Spark core-site.xml file on all nodes:

# vi $SPARK_HOME/yarn-remote-client/core-site.xml

Change sandbox to master:

<?xml version="1.0"?>
<?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
<configuration>
    <property>
        <name>fs.default.name</name>
        <value>hdfs://master:9000</value>
    </property>
    <property>
        <name>dfs.client.use.legacy.blockreader</name>
        <value>true</value>
    </property>
</configuration>

Configuration Spark yarn-site.xml file on all nodes:

# vi $SPARK_HOME/yarn-remote-client/yarn-site.xml

Change sandbox to master:

<configuration>
    <property>
        <name>yarn.resourcemanager.scheduler.address</name>
        <value>master:8030</value>
    </property>
    <property>
        <name>yarn.resourcemanager.address</name>
        <value>master:8032</value>
    </property>
    <property>
        <name>yarn.resourcemanager.webapp.address</name>
        <value>master:8088</value>
    </property>
    <property>
        <name>yarn.resourcemanager.resource-tracker.address</name>
        <value>master:8031</value>
    </property>
    <property>
        <name>yarn.resourcemanager.admin.address</name>
        <value>master:8033</value>
    </property>
    <property>
        <name>yarn.application.classpath</name>
        <value>/usr/local/hadoop/etc/hadoop, /usr/local/hadoop/share/hadoop/common/*, /usr/local/hadoop/share/hadoop/common/lib/*, /usr/local/hadoop/share/hadoop/hdfs/*, /usr/local/hadoop/share/hadoop/hdfs/lib/*, /usr/local/hadoop/share/hadoop/mapreduce/*, /usr/local/hadoop/share/hadoop/mapreduce/lib/*, /usr/local/hadoop/share/hadoop/yarn/*, /usr/local/hadoop/share/hadoop/yarn/lib/*, /usr/local/hadoop/share/spark/*</value>
    </property>
</configuration>

Starting Hadoop multi-node cluster on master node:

# $HADOOP_COMMON_HOME/sbin/start-dfs.sh && $HADOOP_COMMON_HOME/sbin/start-yarn.sh

Track/Monitor/Verify Hadoop cluster on master node:

# jps
1650 SecondaryNameNode
2043 Jps
1484 NameNode
1789 ResourceManager

Track/Monitor/Verify Hadoop cluster on slaves nodes:

# jps
1343 DataNode
1434 NodeManager
1530 Jps

Start Apache Spark on msater node:

# $SPARK_HOME/sbin/start-all.sh

Track/Monitor/Verify Apache Spark cluster on master node:

# jps
2148 Jps
2084 Master
1650 SecondaryNameNode
1484 NameNode
1789 ResourceManage

Track/Monitor/Verify Apache Spark cluster on slaves nodes:

# jps
1733 Jps
1343 DataNode
1434 NodeManager

Monitor Hadoop ResourseManage and Hadoop NameNode via web-version ResourceManager http://master:8088, because I run Docker in Ubuntu server and Ubuntu Server IP address is 172.16.136.128, so I can also visit http://172.16.136.128:8088:

Running Apache Spark on YARN with Docker

Apache Spark Web UI http://master:8080 and I can also visit http://172.16.136.128:8085:

Running Apache Spark on YARN with Docker

Issues

If got following error when start Apache Spark on msater node by $SPARK_HOME/sbin/start-all.sh command:

slave2: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave2.out
slave1: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave1.out
slave2: failed to launch org.apache.spark.deploy.worker.Worker:
slave2:      at java.lang.ClassLoader.loadClass(libgcj.so.10)
slave2:      at gnu.java.lang.MainThread.run(libgcj.so.10)
slave2: full log in /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave2.out
slave1: failed to launch org.apache.spark.deploy.worker.Worker:
slave1:      at java.lang.ClassLoader.loadClass(libgcj.so.10)
slave1:      at gnu.java.lang.MainThread.run(libgcj.so.10)
slave1: full log in /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave1.out

Check OS Java version and list old version Java:

# echo $JAVA_HOME
/usr/java/default
# rpm -qa | grep java
java_cup-0.10k-5.el6.x86_64
gcc-java-4.4.7-16.el6.x86_64
java-1.5.0-gcj-devel-1.5.0.0-29.1.el6.x86_64
R-java-devel-3.2.3-1.el6.x86_64
R-java-3.2.3-1.el6.x86_64
java-1.5.0-gcj-1.5.0.0-29.1.el6.x86_64

List installed JDK:

# rpm -qa | grep jdk
jdk-1.7.0_51-fcs.x86_64

Remove old version Java and JDK:

# rpm -e --nodeps java_cup-0.10k-5.el6.x86_64
# rpm -e --nodeps java-1.5.0-gcj-1.5.0.0-29.1.el6.x86_64

Related Articles

0.00 avg. rating (0% score) - 0 votes