Distributed Machine Learning Studio

Distributed Machine Learning Studio is Drag-and-Drop style ML Plaform that based on Apache Spark computation engine. DMLS supply similar experience as Microsoft Azure Machine Learning Studio and Aliyun PAI, it benifits non-cloud user doing ML task on Big Data.

Instructions of setting up

Initialize metadata store

Get IP address of host

IP_ADDR=$(ifconfig en0 | grep “inet” | awk ‘{ print $2}’ | awk ‘NR==2{print}’) sudo hostname quickstart.cloudera

docker run -itd --name=mysql -p 3306:3306 fluxcapacitor/sql-mysql docker ps -a | grep fluxcapacitor/sql-mysql | awk ‘{print $1}’ | xargs -I {} docker cp dmls_metadata.sql {}:/ docker exec -it $(docker ps -a | grep fluxcapacitor/sql-mysql | awk ‘{print $1}’) bash -c ‘mysql -uroot -ppassword < /dmls_metadata.sql’

Start Web REST project

docker run -itd --name=tomcat --net=host -p 8080:8080 tomcat docker cp dmls-rest.war $(docker ps -a | grep tomcat | awk ‘{print $1}’):/usr/local/tomcat/webapps docker exec -it $(docker ps -a | grep tomcat | awk ‘{print $1}’) bash

Deploy web UI project

docker cp dmls-ui $(docker ps -a | grep tomcat | awk ‘{print $1}’):/usr/local/tomcat/webapps

Start up HDFS

docker run -itd --name=metastore --net=host -e MYSQL_MASTER_SERVICE_HOST=192.168.154.81 -e MYSQL_MASTER_SERVICE_PORT_MYSQL_NATIVE=192.168.154.81 fluxcapacitor/metastore-1.2.1 docker exec -it $(docker ps -a | grep fluxcapacitor/metastore-1.2.1 | awk ‘{print $1}’) bash

Add hostname quickstart.cloudera

docker run --hostname=quickstart.cloudera --privileged=true -t -i -p 127.0.0.1:7180:7180 -p 8020:8020 registry.cn-hangzhou.aliyuncs.com/mysky528/cdh_quickstart /usr/bin/docker-quickstart

Cloudera Manager is not started by default, start it

sh /home/cloudera/cloudera-manager --express