Book Image

Learning CoreOS

By : Kingston Smiler. S, Shantanu Agrawal
Book Image

Learning CoreOS

By: Kingston Smiler. S, Shantanu Agrawal

Overview of this book

CoreOS is an open source operating system developed upon the Linux kernel. The rise of CoreOS is directly related to the rise of Docker (a Linux container management system). It is a minimal operating system layer and takes a different approach to automating the deployment of containers. The major difference between CoreOS and other Linux distributions is that CoreOS was designed to deploy hundreds of servers. CoreOS immensely helps the users to create systems, which are easy to scale and manage, making life easier for all, be it developer, QA, or deployer. This book is all about setting up, deploying, and using CoreOS to manage clusters and clouds. It will help you understand what CoreOS is and its benefits as a cloud orchestration platform. First, we’ll show you how to set up a simple CoreOS instance with single node in the cluster and how to run a Docker container inside the CoreOS instance. Next, you’ll be introduced to Fleet and systemd, and will deploy and distribute Docker services across different nodes in cluster using Fleet. Later, you’ll be briefed about running services in a cluster with constraints, publishing the services already running on the cluster to new services, and making your services interact with each other. We conclude by teaching you about advanced container networking. By the end of the book, you will know the salient features of CoreOS and will be able to deploy, administrate, and secure a CoreOS environment.
Table of Contents (15 chapters)
Learning CoreOS
Credits
About the Authors
About the Reviewer
www.PacktPub.com
Preface
Index

Predefined constraints using metadata


This mechanism enables a service to be runn on a machine having matching metadata configured in the metadata parameter of the coreos.fleet section. Metadata can be used to describe a member properties such as disk type, region, platform, and special member property like exposed public IPs and so on. Since it is provided as a multiple key-value pair, the flexibility it provides is immense for defining a member.

The metadata can then also be used to associate services to be run on those members. For instance, we can say that a particular service is supposed to run on members that are running in a particular region and/or having a particular disk type and/or having a particular member type (bare metal, cloud, and so on) and/or having a particular provider (machine vendor, cloud provider, and so on).

In our example, we will create three members, each having their own metadata, and then bind the service to run on a metadata matching its property. The following...