Cordium documentation · Latest
Quick Installation Guide
On this page +
This is a quick guide for you to install a full-fledged single-node Cordium Cluster on any cheap Linux machine, VM or cloud VPS (e.g. DigitalOcean droplet, Hetzner, AWS EC2, etc...). This single-node Cluster is good enough for development, personal, or undemanding production use cases.
You can also use Octelium Agent Skills with AI coding agents (e.g. Claude Code, Codex, etc...) to help you install Cordium and its underlying Octelium Cluster, create your first Users and Policies, and run your first Workspaces. Simply describe what you want to do, and your agent can prepare the configuration and run the relevant CLIs for you. You can find the installation instructions here or via Ask AI → AI agent skills above.
To install a production, scalable multi-node Cluster over a typical cloud-based or on-premise Kubernetes installation, we recommend referring to this guide here.
Requirements
Cordium uses the exact same installer script used for the Octelium Cluster quick installation guide, which means that it has the same requirements: a freshly installed Linux VM/VPS and a domain or subdomain that you actually own. Since Workspaces are actual development environments that pull images, build code and run language toolchains, we recommend a machine with at least 4 vCPUs, 8 GB of RAM and 80 GB of disk.
Fresh Installation
To install the Cluster, simply SSH into your VPS/VM as the Linux root user and run the installer with the --cordium flag. The installer first installs an Octelium Cluster and then automatically installs the Cordium package at the end:
The --longhorn flag installs Longhorn as the default Kubernetes storage class along with the Kubernetes CSI snapshot controller and a default VolumeSnapshotClass named longhorn-snapshot-vsc. It is optional but highly recommended, since Template pre-builds and WorkspaceSnapshots require a storage backend that supports CSI volume snapshots. Without it, Cordium uses the default local storage of the single-node cluster, and everything else, including persistent Workspaces and EXCLUSIVE Volumes, works normally.
The script should take a few minutes depending on your VM's capabilities to finish.
It's absolutely recommended to read the Octelium Cluster quick installation guide to understand how to set up your DNS records and how to initialize and log in to the Cluster correctly.
Installation on an Existing Octelium Cluster
If you already have a running Octelium Cluster that was installed via the quick installation guide, you only need to label the Kubernetes node to allow it to host Workspaces as follows:
And then install the Cordium package as follows:
Install the CLIs
Now from your own machine, install the cordium CLI as well as the Octelium CLIs, which are used to log in to the Cluster, manage the Octelium resources and connect to the Cluster (read more here):
What Now?
Log in to the Cluster as shown to you at the end of the installation and run your first Workspace:
You can also visit the Cordium web portal via your web browser at https://cordium.<DOMAIN> to create Workspaces, open browser-based terminals, manage Spaces and Templates and chat with the Cordium Agent.
To give your teammates access to Cordium, add them as Octelium Users and authorize them via a Policy as shown here. We also recommend following the first steps guide to get a practical overview of Templates, secretless access and snapshots.