2.1. Download the new software onto each node in the cluster

The new version will reside alongside the existing version.

On each node of the cluster, download and set up the repo file for the new version. Confirm that the HDP repository is configured.

  1. For RHEL, CentOS, or Oracle Linux, run the following commands.

    wget http://public-repo-1.hortonworks.com/HDP/<os>/2.x/updates/2.2.6.0/hdp.repo

    where <os> is centos5 or centos6.

    cp hdp.repo /etc/yum.repos.d/

    yum repolist

  2. For SLES:

    wget http://public-repo-1.hortonworks.com/HDP/<os>/2.x/updates/2.2.6.0/hdp.repo

    where <os> is suse11sp3 or sles11sp1.

    cp hdp.repo /etc/zypp/repos.d

    zypper repos

  3. For Ubuntu or Debian:

    wget http://public-repo-1.hortonworks.com/HDP/<os>/2.x/updates/2.2.6.0/hdp.list

    where <os> is ubuntu12 or debian6.

    cp hdp.list /etc/apt/sources.list.d

    apt-get list

Install the new version of software on each node, based on the component topology of your cluster.

For example, some nodes will have core Hadoop components; others will have Zookeeper processes. If only one node has the Hive Server deployed and running, install the new version of Hive on that node. Include the new version number.

  1. For RHEL, CentOS, or Oracle Linux:

    yum install hive_2_2_6_0_2800*

  2. For SLES:

    zypper install hive_2_2_6_0_2800*

  3. For Ubuntu and Debian:

    apt-get install 'hive_2_2_6_0_2800*'

Appendix B contains a partial list of HDP 2.2.6 package names.

At the end of this process your cluster will still be running your current software. The new software will be stored in a new directory on each node. You will see the following directories in /usr/hdp:

2.2.0.0-2041 2.2.6.0-2800 current


loading table of contents...