If your cluster does not have access to the Internet, or you are creating a large cluster and you want to conserve bandwidth, you need to provide access to the bits using an alternative method.
Set up the local mirror repositories as needed for HDP and HDP Utils.
For more information on your options, see Deploying HDP In Production Data Centers with Firewalls.
From the Ambari Server host, fetch the Ambari repository file or RPM package as described in Set Up the Bits. You need a connection to the Internet for this step.
If you do not have a connection to the Internet for this machine, you should follow the instructions in Deploying HDP In Production Data Centers with Firewalls and be sure to perform the optional steps for setting up the Ambari local repository.
Configure Ambari Server so that it knows to connect to the mirrored repositories during installation.
On Ambari Server, browse to the stacks definitions directory
cd /var/lib/ambari-server/resources/stacks
There are two stack definitions in this directory:
HDP
andHDPLocal
. The HDP definition points to the publicly hosted HDP software packages. You must modify theHDPLocal
definition to point to the local repositories you have set up.Browse to the stack
HDPLocal/
directory. For example, for the 1.3.7 version of the stack:$stack-version
/reposcd HDPLocal/1.3.7/repos
Use a text editor to edit the repo info file. For example:
vi repoinfo.xml
You must update the
<baseurl>
value to point to your local repositories for each operating system that your cluster includes. So, for example, if your system includes hosts running CentOS 6, to point to the HDP 1.3.7 repositories, you would update stanzas to look something like this:<os type="centos6"> <repo> <baseurl>http://{your.hosted.local.repository}/HDP-1.3.7/repos/centos6</baseurl> <repoid>HDP-1.3.7</repoid> <reponame>HDP</reponame> </repo> </os>
The appropriate relative path depends on how you have set up your local repos.
Important If you have mixed operating systems in your cluster (for example, CentOS 6 and RHEL 6), you must configure the repositories and have properly edited <os type> stanzas for both OSes - centos6 and redhat6. If you do not, some hosts in your cluster will not be able to retrieve the software packages for their operating system.
Save this file.
Configure which JDK to use and how the JDK will be downloaded and installed.
If you have not already installed the JDK on all hosts, and plan to use Oracle JDK 1.6, download jdk-6u31-linux-x64.bin to
/var/lib/ambari-server/resources.
If you plan to use a JDK other than Oracle JDK 1.6, you must install the JDK on each host in your cluster and use the -j flag when running
ambari-server setup
. See JDK Requirements for more information on supported JDKs.If you have already installed the JDK on all your hosts, you must use the -j flag when running
ambari-server setup
Note See Setup Options for more information on the -j flag.