1.下载PIO源码&安装:
https://github.com/apache/incubator-predictionio/tree/v0.9.1
$ tar zxvf incubator-predictionio-0.9.1.zip
$ cd incubator-predictionio-0.9.1
$ ./make-distribution.sh
$ tar zxvf PredictionIO-0.9.1.tar.gz
$ mkdir PredictionIO-0.9.1/vendors
2.下载Spark1.2 & ES1.4.4 & Hbase1.0
http://d3kbcqa49mib13.cloudfront.net/spark-1.2.0-bin-hadoop2.4.tgz
https://download.elasticsearch.org/elasticsearch/elasticsearch/elasticsearch-1.4.4.tar.gz
http://archive.apache.org/dist/hbase/hbase-1.0.0/hbase-1.0.0-bin.tar.gz
解压到 vendors目录 , tar zxvf xxxx.tar.gz
3.配置Hbase:
3.1 vendors/hbase-1.0.0/conf/hbase-site.xml,加入如下配置
hbase.rootdir
file:///~xxx/vendors/hbase-1.0.0/data
hbase.zookeeper.property.dataDir
/~xxx/vendors/hbase-1.0.0/zookeeper
3.2 vendors/hbase-1.0.0/conf/hbase-env.sh,配置JAVA环境变量
export JAVA_HOME=/usr/lib/jvm/java8
3.3 vendors/elasticsearch-1.4.4/config/elasticsearch.yml ,设置cluster.name
cluster.name: elasticsearch
4.配置PIO
PredictionIO-0.9.1/conf/pio-env.sh,配置如下:
# Elasticsearch Example
PIO_STORAGE_SOURCES_ELASTICSEARCH_TYPE=elasticsearch
PIO_STORAGE_SOURCES_ELASTICSEARCH_HOSTS=localhost
PIO_STORAGE_SOURCES_ELASTICSEARCH_PORTS=9300
PIO_STORAGE_SOURCES_ELASTICSEARCH_HOME=$PIO_HOME/vendors/elasticsearch-1.4.4
PIO_STORAGE_SOURCES_ELASTICSEARCH_CLUSTERNAME=elasticsearch
# HBase Example
PIO_STORAGE_SOURCES_HBASE_TYPE=hbase
PIO_STORAGE_SOURCES_HBASE_HOSTS=0
PIO_STORAGE_SOURCES_HBASE_PORTS=0
PIO_STORAGE_SOURCES_HBASE_HOME=$PIO_HOME/vendors/hbase-1.0.0
5. 配置PIO环境变量,方便使用命令行
.bash_profile 加入如下:
export PIO091_HOME=/Users/gongsy/predictionio-0.9.1/PredictionIO-0.9.1/bin;
6.启动PIO 以及实用命令
$PIO091_HOME/pio-start-all 启动
$PIO091_HOME/pio-stop-all 停止
$PIO091_HOME/pio status 查看状态
$PIO091_HOME/pio app new xxxxApp
$PIO091_HOME/pio app list
$ cd xxxxEngine 到安装的engine目录下
$PIO091_HOME/pio build --verbose
$PIO091_HOME/pio train
$PIO091_HOME/pio deploy
注意 pio build时 ,sbt-launch-0.13.7.jar文件生成错误:
/Users/gongsy/predictionio-0.9.1/PredictionIO-0.9.1/sbt/sbt-launch-0.13.7.jar