24 Feb 2025

Rancher and Galera Cluster in the partner repository

If you’ve ever wanted to deploy Galera Cluster on Rancher by SUSE, you now can since Galera Cluster is a partner chart in their repository. This means Codership’s Galera Cluster is a SUSE Ready Verified partner. A direct link to the HELM charts for MySQL/Galera Cluster is here.

What is Rancher? Rancher is a Kubernetes management tool to deploy and run clusters anywhere and on any provider. Rancher can provision Kubernetes from a hosted provider, provision compute nodes and then install Kubernetes onto them, or import existing Kubernetes clusters running anywhere.

Rancher comes with very good documentation. We focus on deploying this on DigitalOcean and you get started by simply doing a:

git clone https://github.com/rancher/quickstart

You’ll notice from there that you have the ability to deploy on many platforms: Amazon Web Services, Microsoft Azure Cloud, DigitalOcean, Google Cloud Platform, Harvester, Hetzner Cloud, Linode, Scaleway, and Outscale. For what it is worth, you can also run this locally with Vagrant and VirtualBox, but it takes a good 6GB of RAM for Kubernetes, and Galera Cluster ideally needs 2GB RAM per node.

Once cloned, go to the Terraform directory (quickstart/rancher/do), and rename the terraform.tfvars.example to terraform.tfvars. Yes, you are configuring this with Terraform. Edit it to include your DigitalOcean access key (this is very similar to what we utilise in Galera Manager), then create a rancher_server_admin_password. You can optionally specify region/prefixes/droplet size (remember, use something large enough); this is however optional. Now run terraform init, and then do terraform apply --auto-approve. Now you can use the id_rsa key to SSH into the Rancher Server (it is located in quickstart/rancher/do).

Now you have 2 Kubernetes clusters, ready to go. And now you can access the partner charts. You do not have to do this via a custom Helm chart repository either.

If you fancy doing it manually, instructions are part of the chart. We do recommend 4GB of RAM, and 16GB of storage at the bare minimum. We have also documented how you can connect, and even force a bootstrap of a node from a particular pod.

In conclusion, this is yet another way for you to get access to our Helm charts. Today the latest release is for 8.0.40, but we expect that another release will be out fairly shortly.

Happy Kubernetes clustering with Galera Cluster!