通过Ubuntu 14.04手动安装Hadoop多节点集群

Question

I am a newcomer to Hadoop. For my College project we are given 4 VMs. I need to configure a multi-mode Hadoop cluster on this ( 1 master 3 slaves) and run my webapp on it. I would be using HBase in my project. Usually CentOS is used for installation and deployment of HDP, whereas I was given ubuntu. I cannot use Apache ambari plugin for installation as it is not supported in Ubuntu. I need to manually deploy them, Hence I tried looking out for tutorials. I looked out for a tutorial to install HDP multinode clusters on ubuntu and found this [ http://www.michael-noll.com/tutorials/running-hadoop-on-ubuntu-linux-multi-node-cluster/] But its too outdated (2010)
I have the official documentation here, but I am not able to follow it properly.
[ http://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.1-latest/bk_installing_manually_book/content/rpm-chap2-3.html] and I tried following them.

Could someone suggest me somelinks which are latest, a tutorial with decent amount of screenshots for installation of multinode clusters over Ubuntu 14.04 ( 12.04 is also fine).

Thanks a lot!!

Answer 1

The Michael Noll tutorial is too old, I think. I found this site:

• https://www.digitalocean.com/community/tutorials/how-to-install-hadoop-on-ubuntu-13-10

I have a mini cluster (with 5 slaves and a master) in my University Lab. Ubuntu 12.04 and Hadoop 2.5.0 is there. Furthermore, I have a VM cluster in my laptop (2 slaves and a master) of Hadoop 1.2.1 on Ubuntu 12.04 too.

But I couldn't install Hadoop (any version) in Ubuntu 14.04. I don't remember the cause, but I think it was some problem with Java version (I don't check that).

I hope that help you!

Answer 2

I can across the same issue to install HDP 2.2 on Ubuntu 14.04, and found a solution. I documented everything here: http://www.swiss-scalability.com/2014/12/install-hdp-22-on-ubuntu-1404-trusty.html

In a nutshell, the magic happens here:

sed -e "s/14.04/12.04/g" -i /etc/*-release

And the you can install or restart ambari-agent, it will be able to communicate with ambari-server.