Book Image

Apache Mesos Essentials

By : Dharmesh Kakadia
Book Image

Apache Mesos Essentials

By: Dharmesh Kakadia

Overview of this book

<p>Apache Mesos is a cluster manager that provides efficient resource isolation and sharing across distributed applications, or frameworks. It allows developers to concurrently run the likes of Hadoop, Spark, Storm, and other applications on a dynamically shared pool of nodes. With Mesos, you have the power to manage a wide range of resources in a multi-tenant environment.</p> <p>Starting with the basics, this book will give you an insight into all the features that Mesos has to offer. You will first learn how to set up Mesos in various environments from data centers to the cloud. You will then learn how to implement self-managed Platform as a Service environment with Mesos using various service schedulers, such as Chronos, Aurora, and Marathon. You will then delve into the depths of Mesos fundamentals and learn how to build distributed applications using Mesos primitives.</p> <p>Finally, you will round things off by covering the operational aspects of Mesos including logging, monitoring, high availability, and recovery.</p>
Table of Contents (15 chapters)
Apache Mesos Essentials
Credits
About the Author
About the Reviewers
www.PacktPub.com
Preface
Index

High availability


Most businesses depend on the availability of the infrastructure for continuous business operations. Mesos as a data center kernel not only provides high availability to the applications, but also has several features to ensure that various Mesos components are resilient to failures.

Master high availability

Mesos high availability depends critically on the availability of the Mesos master. As noted previously, Mesos uses ZooKeeper to ensure that one Mesos master is always available even in face of failures. The ZooKeeper elects the acting master, and the other master will be on standby. We have seen how to run Mesos with ZooKeeper in Chapter 6, Understanding Mesos Internals.

ZooKeeper configuration, at the minimum, should set the dataDir. Its value must be set to the persistent storage directory and never to the default value /tmp given in the sample ZooKeeper configuration file. In multiserver settings, a ZooKeeper instance determines its identity reading the myid field...