Based on: http://www.eligotech.com/using-vagrant-for-setting-up-a-cloudera-based-hadoop-cluster/
This project uses Vagrant to create three virtual machines. Then you can connect to Cloudera Manager to install Hadoop and other tools.
- Install Vagrant
- Install git
- Clone this project.
- Connect to project directory.
- Run 'vagrant up'
- Visit http://10.211.55.100:7180 for the Cloudera Manager utility.
- Use admin/admin to login.
- Select Cloudera Standard to use free licenses, then click continue.
- Click continue.
- Enter the following IP addresses, one per line, then click search. 10.211.55.100 10.211.55.101 10.211.55.102
- Click continue.
- Unselect SOLR and Impala (or don't) then click continue. (If you following this advice, you can't select 'All Services' in step 20 below)
- Select 'another user' then enter 'vagrant' as the username.
- Select the 'vagrant' file in this project as the private key file.
- Click continue.
- Click ok to continue with no passphrase.
- Click continue after installation is complete on all three nodes.
- Click continue after parcels are installed.
- Click continue after hosts are inspected.
- Select HDFS, MapReduce, ZooKeeper, HBase, Hive, Oozie, Hue, and Sqoop.
- Select 'Inspect Role Assignments'
- Ensure Zookeeper is on 'vm-cluster-node1' (and only there)
- Ensure NameNode is on 'vm-cluster-node1'.
- Ensure HBase Master is on 'vm-cluster-node1'.
- Click continue.
- Click 'Test Connection'
- Click continue when all connections have been successful.
- Click continue after reviewing the configuration settings.
- Click continue after cluster services have started.
- You're Done!
Handy URLS
CLOUDERA MANAGER: http://10.211.55.100:7180/ NAME NODE http://10.211.55.100:50070/dfshealth.jsp JOB TRACKER: http://10.211.55.101:50030/jobtracker.jsp