Skip to content

exact-payments/kubernetes-ops

 
 

Repository files navigation

ToC

Table of contents generated with markdown-toc

What is kubernetes-ops

TL;DR kubernetes-ops is a way to run Kubernetes and applications on it in a GitOps way using Terraform and Github Actions. We have tried to minimuze the number of tools sets used to keep what is already complex from being even more complex.

The long version

Kubernetes Ops has been a culmination of how we have been helping clients use Kubernetes over the years. There has been a lot of trial an error as we have grown up with Kubernetes. This represents how we are currently helping our clients use Kubernetes and how we help them maintain their infrastructure. You can view this as a reference implementation of a fully productionalized Kubernetes setup.

We lean towards the immutable infrastructure and Gitops Flow methodologies and use no configuration management tools. Everything starts out in Git as either code or configuration. Items are manipulated to what we want the desired state to be and that is applied onto the infrastructure.

One of the biggest problem that this repository helps out with is what people are starting to refer to as "day 2" problems (where "day 1" is creation). The "day 1" problems are well documented and there are plenty of tutorials out there for it. The problem with these tutorials and examples are that they mostly leave you hanging on how to move forward with the infrastructure pieces. Day 2 problems are: what is the upgrade, patching, and modification strategy, how do I manage the infrastructure git repository, etc?

This is where we think we can provide some contribution. With our experience in managing many Kubernetes clusters over the years, we think we can provide this information. Creation of your cluster is about 10 to maybe 20 percent of the infrastructure activity (if that), making changes to the infrastructure to suit your needs as time moves on is the bulk of the activity and finally deletion of the entire or parts of the infrastructure as new items comes into play.

If you follow through the instructions, you might think this is overly complex. If this is your first time playing around with Kubernetes, it probably is overly complex and this project is probably not well suited for you at this time. When you want to take Kubernetes into production, this is where we think this set of methodologies starts to shine. For example, this gives you an example of how to lay out a repository for you infrastructure. It gives you the process and workflow to create and update infrastructure pieces. From working with many clients, we have come to a place where managing the infrastructure in this way has made sense and has worked out really well in large and small teams.

The stack this will create

the stack

Whys

Why do all of this?

Isn't there already projects that bring up Kubernetes for me? Why don't I just use GKE, EKS, AKS, *KS?

Yes, there is and we use all of that. You can use anyone of those services and go to the respective web console and bring up a Kubernetes cluster. In our opinion and from our experience this is fine if you are testing out Kubernetes or just trying out something new. It is fast and easy to understand what is going on. However, when you want to bring that "new thing" into production, managing it that way is not ideal. It is hard to reproduce from dev to qa to prod. Making manual changes are hard to track and very error prone.

We stress that this project does not represent the "easy" way of managing infrastructure. This project represents a way to manage infrastructure in a Gitops flow kinda way and in a sane way where a team of people can work on it together.

GitOps workflow, how does that play into this project?

This project mainly follow a Gitops workflow methodology. Changes are made in this repository to code or configs in a branch. A PR can be opened on that branch where other team members can review the changes. Then depending on your merging techniques and automation it can be applied or merged then applied to any one environment.

Why an entire repository?

We have found that having an "infrastructure" repository makes sense. You need these items to live somewhere. It is usually not application code and it is an entity all to itself. The infrastructure repository also usually gets fairly large overtime as new items gets added into the software stack and new requirements for services comes along.

As an organization grows, it also tends to be a different set of people that maintains the infrastructure and this repository. You have application developers and DevOps or infrastructure groups. Even if you have those two teams in the same group having this separate is a good delineation on what is actually being changed. If items in here are changed, it is clearly an infrastructure related item.

What tools do we use

We mainly only use open source tools. There might be some paid tools eventually ending up in this repository and we will explicitly label those.

Infrastructure building:

  • Terraform

Kubernetes clusters:

  • GKE
  • EKS

Kubernetes tools:

  • Helm
  • Helm Charts from their repository

Supported built in services

These are the list of services that are maintained for each cloud

Service Name Supported in AWS Supported in GCP source
cert-manager yes yes helm/stable
cluster-autoscaler yes no helm/stable
external-dns yes yes helm/stable
graylog yes yes helm/stable
jenkins yes yes helm/stable
kube-bench yes yes helm/stable
kube-downscaler yes yes helm/stable
loki yes yes loki
nginx-ingress yes yes helm/stable
prometheus blackbox exporter yes yes helm/stable
prometheus operator yes yes helm/stable
sumologic-fluentd yes yes helm/stable
threatstack yes yes Threatstack
helm tiller -rbac enabled yes yes -
vault-helm yes yes Hashicorp

Who is using this

Parsable.com
up.audio
karunalabs.com
Many more!

If you too are using kubernetes-common-services; please submit a PR to add your organization to the list!

Topology

The AWS EKS topology

aws kops topology

  • A very isolated VPC with only a few public IP address exposed to the internet
  • Dedicated subnets for each item types. This allows you to segregate items better.
  • Redundant Kubernetes worker nodes in 3 availability zones

GCP GKE

Kubernetes on GCP via GKE clusters

aws kops topology

  • A very isolated VPC with only a few public IP address exposed to the internet
  • Dedicated subnets for each item types. This allows you to segregate items better.
  • Redundant Kubernetes masters in 3 availability zones
  • Redundant Kubernetes worker nodes in 3 availability zones

How do I start using this?

There are various docs and guides in the docs directory.

Read the setup

This is the first thing you should read. This has all of the setup information that you will need to get started.

main doc

This doc will walk you through setting up a Kubernetes EKS infrastructure on AWS: docs/terraform-github-action-pipeline.md

Need a DevOps consultant?

ManagedKube is a boutique DevOps consulting firm that helps companies run large-scale, reliable applications in a GitOps workflow.

We work side-by-side with our client's development team to architect, design, build, optimize, and operate infrastructure in the cloud (AWS and GCP).

We specialize in Docker/Kubernetes containerized infrastructure.

Check us out at: https://managedkube.com/

Or email us at: support@managedkube.com

License

Copyright 2019 ManagedKube
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
   http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

About

[ManagedKube] Running Kubernetes in production

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • HCL 81.4%
  • Shell 8.2%
  • Smarty 5.1%
  • Makefile 3.0%
  • Go 1.6%
  • Dockerfile 0.7%