Rancher Kubernetes on KVM

This repo contains scripts that will allow you to quickly deploy and test Rancher for POC.

Forked from https://github.com/rancher/quickstart

Summary

The repository contains terraform code to stand up a single Rancher server instance with a 3 node cluster attached to it.

This terraform setup will:

• Create a libvirt guest running rancher/rancher version specified in rancher_version
• Create a custom cluster called cluster_name
• Start count_agent_all_nodes amount of droplets and add them to the custom cluster with all roles

Note: This project is customized for KVM servers running Openvswitch. Installation of these dependencies can be complex and is outside the scope of this project.

It has also been customized for use with the Terraform PowerDNS provider.

• https://www.terraform.io/docs/providers/powerdns/index.html

Prereqs

KVM Server running Openvswitch

• https://github.com/mrlesmithjr/ansible-kvm
• https://docs.openvswitch.org/en/latest/intro/install/distributions/

PowerDNS Authoritative Server with API Access

• https://github.com/2stacks/terraform-powerdns

Terraform and the terraform-provider-libvirt

• https://www.terraform.io/downloads.html
• https://github.com/dmacvicar/terraform-provider-libvirt#installing

How to use

• Clone this repository and go into the rancher-kvm folder
• Move the file secret.auto.tfvars.example to secret.auto.tfvars and edit.
• Run terraform init
• Run terraform plan
• Run terraform apply -parallelism=1

Note: Adding -parallelism=1 to terraform apply is needed at this time due to an issue with libvirt failing to create multiple seed images in parallel.

When provisioning has finished you will be given the url to connect to the Rancher Server

How to Remove

To remove the VM's that have been deployed run terraform destroy --force

Support for terraform workspace

• Optionally create a new git branch with git checkout -b dev
• Create the new Terraform workspace with terraform workspace new dev
• Initialize the new environment with terraform init

At this point you should change variables as needed. At a minimum you should probably change the prefix and svr_name variable to prevent resource collisions. You can add those variables to dev.tfvars file and then run;

• terraform plan -var-file=dev.tfvars
• terraform apply -var-file=dev.tfvars -parallelism=1

TODO

Optional adding nodes per role

• Start count_agent_etcd_nodes amount of droplets and add them to the custom cluster with etcd role
• Start count_agent_controlplane_nodes amount of droplets and add them to the custom cluster with controlplane role
• Start count_agent_worker_nodes amount of droplets and add them to the custom cluster with worker role

Note: Only adding nodes with all roles is currently supported.