Modelplane Modelplane docs
Version
main Latest

Installation

On this page

The control plane is where everything in Modelplane runs. In this step you’ll install it on a local kind cluster, using Crossplane for reconciliation and the Modelplane APIs. No cloud yet, that comes next.

This step takes about five minutes.

Prerequisites

Install kind, kubectl, and Helm on your machine.

Note
You can run your Modelplane control plane anywhere. This tour uses kind for illustration.

Install the control plane

Crossplane provides the reconciliation engine and package management. Create the kind cluster and install it with Helm:

bash
kind create cluster --name modelplane
bash
helm repo add crossplane-stable https://charts.crossplane.io/stable
helm repo update crossplane-stable
helm install crossplane crossplane-stable/crossplane \
  --namespace crossplane-system --create-namespace \
  --set "args={--enable-dependency-version-upgrades}" \
  --wait

Apply the bootstrap resources. They grant Crossplane the permissions it needs to manage your cluster:

shell
kubectl apply -f https://v0-1.docs.modelplane.ai/examples/getting-started/prerequisites.yaml

prerequisites.yaml
# Modelplane prerequisites. Apply once after installing Crossplane.
#
# These resources grant Crossplane and provider-helm the permissions
# needed to compose Gateway API, MetalLB, and Service/EndpointSlice
# routing resources. They cannot be self-composed because Crossplane
# needs the permissions before it can compose anything.
---
apiVersion: v1
kind: Namespace
metadata:
  name: modelplane-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: crossplane-compose-modelplane
  labels:
    rbac.crossplane.io/aggregate-to-crossplane: "true"
rules:
- apiGroups: [""]
  resources: ["namespaces"]
  verbs: ["*"]
# Selectorless Service plus EndpointSlice composed by ModelEndpoint to route
# the control plane gateway to a remote model endpoint.
- apiGroups: [""]
  resources: ["services"]
  verbs: ["*"]
- apiGroups: ["discovery.k8s.io"]
  resources: ["endpointslices"]
  verbs: ["*"]
- apiGroups: ["gateway.networking.k8s.io"]
  resources: ["gateways", "gatewayclasses", "httproutes"]
  verbs: ["*"]
- apiGroups: ["gateway.envoyproxy.io"]
  resources: ["backends"]
  verbs: ["*"]
- apiGroups: ["metallb.io"]
  resources: ["ipaddresspools", "l2advertisements"]
  verbs: ["*"]
- apiGroups: ["protection.crossplane.io"]
  resources: ["usages"]
  verbs: ["*"]
---
# Give provider-helm a deterministic SA name so we can grant it
# permissions. Without this, the SA name has a random hash.
apiVersion: pkg.crossplane.io/v1beta1
kind: DeploymentRuntimeConfig
metadata:
  name: provider-helm-modelplane
spec:
  serviceAccountTemplate:
    metadata:
      name: provider-helm-modelplane
---
# Grant provider-helm cluster-admin. It installs full Helm charts
# (MetalLB, Envoy Gateway, LeaderWorkerSet, etc.) that create arbitrary
# resource types across namespaces.
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: provider-helm-modelplane
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: cluster-admin
subjects:
- kind: ServiceAccount
  name: provider-helm-modelplane
  namespace: crossplane-system
---
# Apply the DRC to provider-helm automatically via ImageConfig.
apiVersion: pkg.crossplane.io/v1beta1
kind: ImageConfig
metadata:
  name: provider-helm-modelplane
spec:
  matchImages:
  - type: Prefix
    prefix: xpkg.upbound.io/upbound/provider-helm
  runtime:
    configRef:
      name: provider-helm-modelplane

Install Modelplane

The Modelplane Configuration adds the Modelplane APIs and the composition functions that reconcile them:

configuration.yaml
# The Modelplane Crossplane Configuration. Installing it adds the Modelplane
# APIs (InferenceGateway, InferenceClass, InferenceCluster, ModelDeployment,
# ModelCache, ModelService) and the composition functions that reconcile them.
apiVersion: pkg.crossplane.io/v1
kind: Configuration
metadata:
  name: modelplane
spec:
  package: xpkg.upbound.io/modelplane/modelplane:v0.1.0

Wait until the configuration is healthy:

bash
kubectl wait configuration/modelplane --for=condition=Healthy --timeout=5m

Next step

The control plane is running but has nothing to schedule against yet. In the next step, you’ll build the platform to provision a GPU cluster and publish what hardware it offers.