Puppet Class: profile::hadoop::worker::clients
- Defined in:
- modules/profile/manifests/hadoop/worker/clients.pp
Overview
SPDX-License-Identifier: Apache-2.0
Class profile::hadoop::worker::clients
Configure a Analytics Hadoop worker node with extra client tools to connect to Hive and use Sqoop/Spark2/etc..
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 |
# File 'modules/profile/manifests/hadoop/worker/clients.pp', line 7
class profile::hadoop::worker::clients {
# hive::client is nice to have for jobs launched
# from random worker nodes as app masters so they
# have access to hive-site.xml and other hive jars.
# This installs hive-hcatalog package on worker nodes to get
# hcatalog jars, including Hive JsonSerde for using
# JSON backed Hive tables.
include ::profile::hive::client
# We want to exclude spark2 from bullseye installs
if debian::codename::lt('bullseye') {
# Spark 2 is manually packaged by us, it is not part of CDH.
include ::profile::hadoop::spark2
}
# Spark 3 is provided in our custom conda-analytics package
# via pyspark installed in the conda environment in /opt/conda-analytics.
include ::profile::hadoop::spark3
# sqoop needs to be on worker nodes if Oozie is to
# launch sqoop jobs.
class { '::bigtop::sqoop': }
}
|