Book Image

Argo CD in Practice

By : Liviu Costea, Spiros Economakis
Book Image

Argo CD in Practice

By: Liviu Costea, Spiros Economakis

Overview of this book

GitOps follows the practices of infrastructure as code (IaC), allowing developers to use their day-to-day tools and practices such as source control and pull requests to manage apps. With this book, you’ll understand how to apply GitOps bootstrap clusters in a repeatable manner, build CD pipelines for cloud-native apps running on Kubernetes, and minimize the failure of deployments. You’ll start by installing Argo CD in a cluster, setting up user access using single sign-on, performing declarative configuration changes, and enabling observability and disaster recovery. Once you have a production-ready setup of Argo CD, you’ll explore how CD pipelines can be built using the pull method, how that increases security, and how the reconciliation process occurs when multi-cluster scenarios are involved. Next, you’ll go through the common troubleshooting scenarios, from installation to day-to-day operations, and learn how performance can be improved. Later, you’ll explore the tools that can be used to parse the YAML you write for deploying apps. You can then check if it is valid for new versions of Kubernetes, verify if it has any security or compliance misconfigurations, and that it follows the best practices for cloud-native apps running on Kubernetes. By the end of this book, you’ll be able to build a real-world CD pipeline using Argo CD.
Table of Contents (15 chapters)
1
Part 1: The Fundamentals of GitOps and Argo CD
4
Part 2: Argo CD as a Site Reliability Engineer
7
Part 3: Argo CD in Production

Setting up an HA installation

Since we have already used the HA option with Kustomize, let’s see what components are installed, how they handle the HA part, and if there is anything else we can do:

  • API server: This handles all the exterior interaction, so if you are using the CLI or the UI or creating a client, you will communicate with the API. The HA manifests already set two instances for this pod.
  • Repository server: This is responsible for creating the final manifests to apply to the cluster; manifests generation is complicated because of all the templating supported by Argo CD, such as Helm 2 or 3, Kustomize, and Jsonnet. The HA manifests come with two replicas.
  • Application controller: This is where the work is initiated, where the control loop is implemented, and application sync takes place. Initially, you were only able to have one instance, but now, you can have one instance per cluster shard. The HA manifests use one instance of the controller.
  • ...