Introduction

PackOne is used to simplify big data cluster deployment on the clouds include OpenStack, H3CloudOS, EVCloud and CSTCloud etc. It can bootstrap a cluster from scratch in a few clicks, and materialize the cluster into cloud images to boost following cluster creating procedures. Besides, PackOne can scale-out/in clusters by only one click.

This software is inspired by the "serverless" trend in cloud computing and big data processing, with the ambitions to bridge the IaaS to Apache Ambari seamlessly and coordinate Ambari Services into an elastic high-level workspace. Currently, It support to deploy and scale out Hadoop, Spark, Hive, Neo4j, MongoDB, Kylin, Redis and many other clustering software in several clicks or api callings.

Install

1. Choose a linux (Centos 7.5 is verified) host which network can reach the target cloud resource;
2. yum install python36-pip and python36-devel;
3. Create a postgresql db with its information (db_user, db_passwd, db_host, db_port, db_name) collected.

Then run:

pip3.6 install pk1

pip3.6 install -U pip setuptools

pk1 setup --database db_user:db_passwd:db_host:db_port:db_name

Start Service

pk1 start [--listening 127.0.0.1:11001]

Step 1: Add Cloud (OpenStack, for example)

open http://127.0.0.1:11001/clouds/cloud/add/, and fill the form like:

Step 2: Bootstrap an Ambari cluster

open http://127.0.0.1:11001/engines/cluster/add/, and fill the form like:

Step 3: Materialize/Scale clusters

open http://127.0.0.1:11001/engines/cluster/, select the target clusters, and click the following materialize.../scale... link:

Step 4: Boost cluster creation

Similar to Step 2, open http://127.0.0.1:11001/clouds/cloud/add/, but choose a scale whose name without 'boostrap'.

Stop

pk1 stop

Uninstall

pk1 uninstall

Acknowledge

National Key Research Program of China: Scientific Big Data Management System (No.2016YFB1000600)