Skip to content

Latest commit

 

History

History
95 lines (69 loc) · 3.74 KB

README.md

File metadata and controls

95 lines (69 loc) · 3.74 KB
Raw

Operational Overview with Ambari

GOAL - Using the Hortonworks Sandbox, provide an overview of Apache Ambari; especially from the perspective of the Ops perspective of administering a Hadoop cluster

PREREQUISITE - Sandbox Setup

SEE ALSO - Hortonworks Ambari Page

RECORDED DEMO

Ops with Ambari

PRIOR DEMO CLEANUP - None

Ambari's Role

Apache Ambari aids very specifically with the following areas for a Hadoop cluster

  • Installation
  • Upgrades
  • Configuration
  • Expansion
  • Start/Stop Operations
  • Monitoring

Installation

A quick way to get a feel for the LOE involved in manually configuring a Hadoop cluster is to look at the activities required. The Manual Installation Guide devotes over 250 pages to this, while the comparable operations in the Ambari Installation Guide span about 25 pages.

While both approaches require HW & OS prep activities, the Ambari route quickly differs by only needing two quick CLI operations, yum install ambari-server and ambari-sever setup, before focusing on its GUI wizard process to completely install a cluster.

While the Sandbox Setup that we are using already has an installed version of HDP, you can get a sense of this install wizard by starting the process of adding a new service. Click on Actions > Add Service > Accumulo > Next to see something like presented in the Add Service Wizard screenshot.

Be sure to exit out of this wizard to prevent service form attempting to be installed.

Configuration

As visualized in this figurative example, Hadoop installation files get spread across the file system and across all the nodes in the cluster. Installation with Ambari ensures all get to the right place and subsequent configuration changes with Ambari also ensures appropriate, primarily XML, files are accurate updated as well as guides the operator through any possibly restart activities to ensure they are utilized.

For example, search for execution within /etc/hive/conf/hive-site.xml to find the following stanza.

    <property>
      <name>hive.execution.engine</name>
      <value>tez</value>
    </property>

To see how to set that, as well as what values are acceptable just navigate to Services > Hive > Configs

Settings > Optimization > Execution Engine. You can even change it to MapReduce and save a new version to see the visual compare tools like shown below before reverting back to the prior version.

alt text

Expansion

Ambari shines when it is time to grow your cluster with additional nodes. It already has all the bits to distribute and intelligently deploys all required configuration files.

Start/Stop Operations

Very straight-forward to stop and/or start services. The ability to mark something in Maintenance Mode is great to silence alarms about a particular service not being operational.

Monitoring

As you can see from the Dashboard view, monitoring is a key element of Ambari. All of these events can also be integrated into your unified dashboard that spans multiple technologies and vendors.