Skip to content
Copyright © 2022-2026 basebox GmbH, all rights reserved.
Licensed to be used in conjunction with basebox, only.

// installation

Existing cluster: requirements

Path B · Starting with an existing Kubernetes cluster

Applies to

Product: Server · Audience: Platform Operator

You already run a Kubernetes cluster with GPU nodes and want to install basebox on it. This page lists what the cluster must provide, which steps of the bare-metal path you then skip, and where you start. The choice between the paths is described in Installation paths.

What the cluster must provide

  • Kubernetes 1.23+ (verified: 1.33.7; 1.33+ recommended), Helm 3.x
  • Dynamic volume provisioning with a default storage class
  • An ingress controller (nginx annotations are prepared in the chart) and a TLS path: cert-manager with a ClusterIssuer, an existing TLS secret or local trust
  • CloudNativePG operator for the PostgreSQL clusters
  • NVIDIA GPU Operator (or equivalent) on the nodes running inference or service models in GPU mode; configured accordingly for mixed MIG layouts
  • Registry access to gitea.basebox.health from the nodes' container runtime – or mirrored images
  • Sufficient resources: minimum 5+ cores, 12 GB+ RAM, 1+ GPU, 200 GB+; recommended 10+ cores, 32 GB+ RAM, 2+ GPUs, 500 GB+ SSD/NVMe. How much hardware you need for your number of users and models: Sizing.

Check the cluster

kubectl get nodes -o wide
kubectl get pods -A
kubectl get storageclass
kubectl get svc -A | grep -i ingress
kubectl get pods -n cnpg-system
kubectl get pods -n gpu-operator            # on GPU nodes
kubectl describe nodes | grep -A5 "Allocated resources"

Everything Ready/Running, a default storage class, an ingress IP, GPU resources on the GPU nodes – then the cluster meets the requirements.

If something is missing – usually the GPU Operator or CloudNativePG – add it following the respective step of the bare-metal path: NVIDIA / GPU → GPU Operator, Kubernetes → CloudNativePG operator, Kubernetes → Ingress controller.

What you skip

Step of the bare-metal path On path B
1 · Requirements Hardware does not apply; you still make the DNS, TLS and model decisions
2 · Prepare the server Skipped
3 · NVIDIA / GPU Skipped if the GPU Operator is running; check the MIG layout where planned
4 · Kubernetes Skipped if the check above passes
5 · Storage Still read capacity planning and backup targets
6 · Networking Still define hostname, TLS mode and the AISRV ↔ inference boundary

How to continue

From step 7 on, both paths are identical:

  1. 7 · Install basebox – umbrella chart via Helm; all options under Using Helm charts
  2. 8 · Deploy service models
  3. 9 · Connect inference
  4. 10 · Validate installation

Without internet access you need an offline transfer of images and models; agree the procedure with support@basebox.ai.

Next step: 7 · Install basebox