Using Berkelium

Berkelium offers multiple interfaces tailored to different scientific workflows.

1. The Web User Portal

The Berkelium User Portal is the fastest way to:

Monitor live resource consumption (CPU Cores, RAM, GPUs, and active pods).
Provision new namespaces for grants and research groups.
Download customized kubeconfig connection files.
Inspect active pod statuses and decommission stale namespaces.

Namespace Naming Conventions

When provisioning a namespace in the portal:

RFC 1123 Rules: Namespace names must be between 2 and 63 characters in length, contain only lowercase alphanumeric characters (a-z, 0-9) or hyphens (-), and must begin and end with an alphanumeric character.
Reserved Namespaces: Names must not conflict with cluster system namespaces (e.g., kube-*, *-system, *-operator, cert-manager, monitoring, gatekeeper-*, berkelium-*, etc.) or existing namespaces in the cluster.
Examples: bio-genomics, materials-sim, als-beamline, climate-downscaling.

2. Command-Line Tools: `kubectl` & `k9s`

Standard `kubectl` CLI

The standard tool for interacting with the Kubernetes API server. Replace <your-namespace> with your provisioned namespace (e.g. bio-genomics):

# Check current namespace and cluster context
kubectl config current-context

# List running pods in your scientific namespace
kubectl get pods -n <your-namespace>

# Describe pod events and errors
kubectl describe pod <pod-name> -n <your-namespace>

K9s — Interactive Terminal UI (Recommended CLI Tool)

For a much faster, keyboard-driven terminal experience—especially when working over SSH—we recommend K9s (GitHub: derailed/k9s).

K9s provides a real-time curses-style terminal dashboard on top of your ~/.kube/config that continually watches your namespace resources and lets you view logs (l), drop into container shells (s), describe resources (d), edit manifests (e), or delete pods (Ctrl+d) with single keystrokes:

# Install on macOS or Linux via Homebrew
brew install derailed/k9s/k9s

# Launch K9s scoped to your research namespace
k9s -n <your-namespace>

3. Radar — Visual Cluster Management & Control (Recommended GUI)

For researchers who prefer a fast, modern browser interface to visually control and inspect their Kubernetes workloads, we recommend Radar (GitHub: skyhook-io/radar).

Radar is an open-source Kubernetes visibility and management tool that runs locally as a single lightweight binary or kubectl plugin using your existing ~/.kube/config and kubelogin OIDC credentials:

Interactive Resource & Topology Graph: Visualizes live relationships between your Pods, Batch Jobs, Deployments, Services, PersistentVolumeClaims, and HTTPRoute ingresses.
Direct Workload Control: Inspect, edit manifests, restart, scale, or delete pods and jobs in your namespace directly from a clean browser UI.
Live Log Streaming & Diagnostics: Stream real-time container logs, inspect scheduling events, monitor CPU/memory metrics, and browse container filesystems to quickly troubleshoot CrashLoopBackOff or OOMKilled failures.
Built-in AI MCP Server: Includes a local Model Context Protocol (MCP) server so AI coding assistants (such as Claude Desktop or Cursor) can query live cluster state alongside the Berkelium MCP Endpoint.

Installing & Launching Radar

# Install on macOS or Linux via Homebrew
brew install skyhook-io/tap/radar

# Or install via shell script (Linux / macOS)
curl -fsSL https://get.radarhq.io | sh

# Launch the Radar UI in your browser using your current ~/.kube/config
kubectl radar

Learn more at radarhq.io.

4. Headless & Automated CI/CD

For automated pipelines (e.g. beamline data ingestion triggering upon detector acquisition):

Science IT supports Kubernetes Service Accounts with tightly scoped RBAC role bindings.
Submit a consultation ticket at Science IT Support to obtain a machine robot token for your automated acquisition server.