PredictionIO Install Instructions

PredictionIO Setup Guide

This is a pretty much self-noted guide to setting up PredictionIO in a cluster of 3 servers (for this guide in particular).

You can also setup more servers and distribute the services mentioned here differently, but for the scope of this guide I won't explain how to do that, although you might use the references here to guide yourself into doing that.

Requirements: --

Note: In this guide, all servers share all services, except PredictionIO, which runs only under the master server.

If you wanna distribute PIO, you need to setup a load balancer on top of each Eventserver.

Hadoop 2.6.2 (Fully distributed mode)
Spark 1.5.2 (Fully distributed mode)
Elasticsearch 1.7.4 (Multi master cluster)
HBase 1.1.2 (Multi master cluster)
PredictionIO 0.9.6
Universal Recommender Template Engine (Provided by ActionML)

Setup User: --

1.1 Create user for PredictionIO pio in each server

adduser pio # Give it some password

1.2 Give the pio user sudoers permissions

usermod -a -G sudo pio

1.3 Setup paswordless ssh between all servers of the cluster (a.k.a: Add pub key to authorized_keys)

1.4 Modify /etc/hosts file and name each server

Note: Avoid using "localhost" or "127.0.0.1".

# Change IPs where it corresponds.
10.0.0.1 master
10.0.0.2 slave-1
10.0.0.3 slave-2

Download services in all servers: -- Note: Download everything to a temp folder like /tmp/downloads, we will later move them to the final destinations.

2.1 Download Hadoop 2.6.2 (http://www.eu.apache.org/dist/hadoop/common/hadoop-2.6.2/hadoop-2.6.2.tar.gz)

2.2 Download Spark 1.5.2 (http://www.us.apache.org/dist/spark/spark-1.5.2/spark-1.5.2-bin-hadoop2.6.tgz)

2.3 Download Elasticsearch 1.7.4 (https://download.elastic.co/elasticsearch/elasticsearch/elasticsearch-1.7.4.tar.gz)

DON'T USE 2.0 UNTIL PIO WORKS WITH IT. (Pat said there were some issues).

2.4 Download HBase 1.1.2 (https://www.apache.org/dist/hbase/1.1.2/hbase-1.1.2-src.tar.gz)

2.5 Clone PIO Enterprise

git clone https://github.com/actionml/PredictionIO-Enterprise.git predictionio

2.6 Clone Universal Recommender Template

# cd into /tmp/downloads/predictionio/templates (create it if it doesn't exist)
mkdir -p /tmp/downloads/predictionio/templates && cd $_
git clone https://github.com/pferrel/template-scala-parallel-universal-recommendation universal

Setup Java 1.7 or 1.8 (OpenJDK): -- (Using Debian-based distros)

3.1 Install Java.

sudo apt-get install openjdk-7-jdk

3.2 Check which versions of Java are installed and pick one (Ideally OpenJDK, PIO has issues with Oracle Java.)

sudo update-alternatives --config java

3.3 Set JAVA_HOME env var.

Note: Don't include the /bin folder in the route.

vim /etc/environment
export JAVA_HOME=/path/to/open/jdk/jre

Create Folders for services: --

4.1 Create folders in /opt (You might as well place these services wherever your distro recommends, it's up to you to stick to this convention or not)

mkdir /opt/hadoop
mkdir /opt/spark
mkdir /opt/elasticsearch
mkdir /opt/hbase
mkdir /opt/pio

chown pio:pio /opt/hadoop
chown pio:pio /opt/spark
chown pio:pio /opt/elasticsearch
chown pio:pio /opt/hbase
chown pio:pio /opt/pio

Extract Services: --

5.1 Inside the /tmp/downloads folder, extract all downloaded services.

5.2 Move extracted services to their folders

sudo mv /tmp/downloads/hadoop-2.6.2 /opt/hadoop/
sudo mv /tmp/downloads/spark-1.5.2 /opt/spark/
sudo mv /tmp/downloads/elasticsearch-1.7.4 /opt/elasticsearch/
sudo mv /tmp/downloads/hbase-1.1.2 /opt/hbase/
sudo mv /tmp/downloads/predictionio /opt/pio/

5.3 NOTE: Keep version numbers, if we want to upgrade in the future without losing previous versions, we just need to re-symlink.

5.4 Symlink Folders

sudo ln -s /opt/hadoop/hadoop-2.6.2 /usr/local/hadoop
sudo ln -s /opt/spark/spark-1.5.2 /usr/local/sparl
sudo ln -s /opt/elasticsearch/elasticsearch-1.7.4 /usr/local/elasticsearch
sudo ln -s /opt/hbase/hbase-1.1.2 /usr/local/hbase
sudo ln -s /opt/pio/predictionio /usr/local/pio

Setup clusterized services: --

6.1. Setup Hadoop.

Read: http://www.tutorialspoint.com/hadoop/hadoop_multi_node_cluster.htm

Files config:

etc/hadoop/core-site.xml

<configuration>
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://master:9000</value>
    </property>
</configuration>

etc/hadoop/hadoop/hdfs-site.xml

<configuration>
   <property>
      <name>dfs.data.dir</name>
      <value>file:///usr/local/hadoop/dfs/name/data</value>
      <final>true</final>
   </property>

   <property>
      <name>dfs.name.dir</name>
      <value>file:///usr/local/hadoop/dfs/name</value>
      <final>true</final>
   </property>

   <property>
      <name>dfs.replication</name>
      <value>1</value>
   </property>
</configuration>

etc/hadoop/mapred-site.xml

<configuration>
   <property>
      <name>mapred.job.tracker</name>
      <value>master:9001</value>
   </property>
</configuration>

etc/hadoop/masters

master
etc/hadoop/slaves
```
slave-1
slave-2
```

etc/hadoop/hadoop-env.sh

export JAVA_HOME=${JAVA_HOME}
export HADOOP_CONF_DIR=${HADOOP_CONF_DIR:-"/etc/hadoop"}
for f in $HADOOP_HOME/contrib/capacity-scheduler/*.jar; do
  if [ "$HADOOP_CLASSPATH" ]; then
    export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:$f
  else
    export HADOOP_CLASSPATH=$f
  fi
done
export HADOOP_OPTS="$HADOOP_OPTS -Djava.net.preferIPv4Stack=true"
export HADOOP_NAMENODE_OPTS="-Dhadoop.security.logger=${HADOOP_SECURITY_LOGGER:-INFO,RFAS} -Dhdfs.audit.logger=${HDFS_AUDIT_LOGGER:-INFO,NullAppender} $HADOOP_NAMENODE_OPTS"
export HADOOP_DATANODE_OPTS="-Dhadoop.security.logger=ERROR,RFAS $HADOOP_DATANODE_OPTS"
export HADOOP_SECONDARYNAMENODE_OPTS="-Dhadoop.security.logger=${HADOOP_SECURITY_LOGGER:-INFO,RFAS} -Dhdfs.audit.logger=${HDFS_AUDIT_LOGGER:-INFO,NullAppender} $HADOOP_SECONDARYNAMENODE_OPTS"
export HADOOP_NFS3_OPTS="$HADOOP_NFS3_OPTS"
export HADOOP_PORTMAP_OPTS="-Xmx512m $HADOOP_PORTMAP_OPTS"
export HADOOP_CLIENT_OPTS="-Xmx512m $HADOOP_CLIENT_OPTS"
export HADOOP_SECURE_DN_USER=${HADOOP_SECURE_DN_USER}
export HADOOP_SECURE_DN_LOG_DIR=${HADOOP_LOG_DIR}/${HADOOP_HDFS_USER}
export HADOOP_PID_DIR=${HADOOP_PID_DIR}
export HADOOP_SECURE_DN_PID_DIR=${HADOOP_PID_DIR}
export HADOOP_IDENT_STRING=$USER

Format Namenode
```
`bin/hadoop namenode -format`
```
Start just dfs servers (from master)
```
`sbin/start-dfs.sh`
```
Create /hbase and /zookeper folders (from master)

bin/hdfs dfs -mkdir /hbase /zookeeper

NOTES:

Follow instructions in the guide and just set up master and slaves, then start only the hdfs ("hadoop distrubuted file system"), DO NOT DO start-all.sh

6.2. Setup Spark.

Pretty straight forward: http://spark.apache.org/docs/latest/spark-standalone.html
Start all services (from master)

sbin/start-all.sh
Start just master

sbin/start-master.sh
Start just slaves

sbin/start-slave.sh <master-spark-URL>

6.3. Setup Elasticsearch

Change the conf/elasticsearch.yml file to reflect this:

cluster.name: elasticsearch-pio-poc
node.name: "master" # Change to the name of the slave if the server is a slave.
node.master: true # SET TO TRUE ONLY IN MASTER, OTHERS IS FALSE
node.data: true
discovery.zen.ping.multicast.enabled: false
discovery.zen.ping.unicast.hosts: ["master", "slave-1", "slave-2"] # ADD ALL THE SERVERS

6.4. Setup HBase (Prepare for hell)

THIS IS THE BEST GUIDE FOR THIS VERSION, TRUST NO OTHER (This is actually the official guide) https://hbase.apache.org/book.html#quickstart_fully_distributed

Files config:

conf/hbase-site.xml

<configuration>
  <property>
    <name>hbase.rootdir</name>
    <value>hdfs://master:9000/hbase</value>
  </property>

   <property>
    <name>hbase.cluster.distributed</name>
    <value>true</value>
  </property>

  <property>
    <name>hbase.zookeeper.property.dataDir</name>
    <value>hdfs://master:9000/zookeeper</value>
  </property>

  <property>
    <name>hbase.zookeeper.quorum</name>
    <value>master,slave-1,slave-2</value>
  </property>

  <property>
    <name>hbase.zookeeper.property.clientPort</name>
    <value>2181</value>
  </property>
</configuration>

conf/regionservers
```
master
slave-1
slave-2
```
conf/backupmasters

slave-1

conf/hbase-env.sh

export JAVA_HOME=${JAVA_HOME}
export HBASE_OPTS="-XX:+UseConcMarkSweepGC"
export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS -XX:PermSize=128m -XX:MaxPermSize=128m"
export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS -XX:PermSize=128m -XX:MaxPermSize=128m"
export HBASE_PID_DIR=/var/hbase/pids
export HBASE_MANAGES_ZK=true

Start HBase (from master)

bin/start-hbase.sh
NOTE: I strongly recommend setting all these files just in the master conf folder and just copying the whole conf/* folder to the slaves.

Setup PIO --

7.1 Add /usr/local/pio/bin to the PATH

echo "export PATH=$PATH:/usr/local/pio/bin" >> ~/.bashrc

7.2 Start everything and check it works

pio-start-all # Starts basic services.
pio status # Should return all statuses ok.

7.2 Build

pio build

7.3 Create App

pio app new handmade

7.4 Install pip, follow this guide

7.5 Get Python SDK

sudo pip install predictionio

7.6 Test everything works. (Be sure to have all services running)

$ ./usr/local/pio/examples/integration-test

rojosinalma/predictionio_readme.md

PredictionIO Setup Guide

6.1. Setup Hadoop.

6.2. Setup Spark.

6.3. Setup Elasticsearch

6.4. Setup HBase (Prepare for hell)

ar012 commented Jul 1, 2018