The Hadoop Software Upgrade operation uses many of the same steps used by other PUT software related operations. In most cases those steps run exactly the same way and display the same screens. PUT steps not needed for Hadoop upgrades have been removed. The main difference between the Hadoop Software Upgrade operation and other PUT operations is how packages are installed. In the manual Hadoop upgrade procedure, the user must perform a number of manual steps while upgrading some of the Hadoop components. PUT automates the manual process by automatically selecting and grouping packages based on predefined grouping rules so that PUT can run these steps at the right time.
The following 5 package groups are currently defined for Hadoop Upgrades and are upgraded in the order listed. The 3 upgrade scripts required to automate the manual steps are also listed with their respective groups.
Package Groups | Description |
---|---|
HadoopTools | teradata-hadoop-tools only |
HadoopCore | Core Hadoop and HDFS components
|
Hadoop | Remaining Hadoop applications and monitoring software
|
Release | teradata-hdp-release package only (installed to indicate success) |
ExtraPackages | Extra packages that do not require a specific package group |
Each package group upgrades a specific set of rpms so that pre- and post-upgrade scripts can run in between them. Each upgrade script runs a sequence of hcli commands to automate the manual steps. It is recommended that the teradata-hadoop-tools package be upgraded prior to starting a Hadoop upgrade to assure that the latest version of the hcli utility is in place. Alternatively, you can place the latest teradata-hadoop-tools package in the same directory as the other Hadoop packages. When PUT finds a new tools package it installs it first in the HadoopTools package group. Each group is upgraded using the same 3 PUT steps: Hadoop Preinstall, Package Install and Hadoop Post Install, in this order. These steps are repeated for each group until all packages have been upgraded. A confirmation screen is displayed after each set of rpms has been upgraded.