Deployment Workflow

This guide deploys a self-hosted Upbound cluster in AWS.

Disconnected Spaces allows you to host control planes in your preferred environment.

Prerequisites

To get started deploying your own Disconnected Space, you need:

• An Upbound organization account string, provided by your Upbound account representative
• A token.json license, provided by your Upbound account representative

• An AWS account and the AWS CLI

important

Disconnected Spaces are a business critical feature of Upbound and requires a license token to successfully complete the installation. Contact Upbound if you want to try out Upbound with Disconnected Spaces.

Provision the hosting environment

Create a cluster

Configure the name and target region you want the EKS cluster deployed to.

export SPACES_CLUSTER_NAME=upbound-space-quickstart
export SPACES_REGION=us-east-1

Provision a 3-node cluster using eksctl.

cat <<EOF | eksctl create cluster -f -
apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig
metadata:
  name: ${SPACES_CLUSTER_NAME}
  region: ${SPACES_REGION}
  version: "1.29"
managedNodeGroups:
  - name: ng-1
    instanceType: m5.xlarge
    desiredCapacity: 3
    volumeSize: 100
    iam:
      withAddonPolicies:
        ebs: true
iam:
  withOIDC: true
  serviceAccounts:
    - metadata:
        name: aws-load-balancer-controller
        namespace: kube-system
      wellKnownPolicies:
        awsLoadBalancerController: true
    - metadata:
        name: efs-csi-controller-sa
        namespace: kube-system
      wellKnownPolicies:
        efsCSIController: true
addons:
  - name: vpc-cni
    attachPolicyARNs:
      - arn:aws:iam::aws:policy/AmazonEKS_CNI_Policy
  - name: aws-ebs-csi-driver
    wellKnownPolicies:
      ebsCSIController: true
EOF

Configure the pre-install

Set your Upbound organization account details

Set your Upbound organization account string as an environment variable for use in future steps

export UPBOUND_ACCOUNT=<your-upbound-org>

Set up pre-install configurations

Export the path of the license token JSON file provided by your Upbound account representative.

# Change the path to where you saved the token.
export SPACES_TOKEN_PATH="/path/to/token.json"

Set the version of Spaces software you want to install.

export SPACES_VERSION=<!-- spaces_version -->

Set the router host and cluster type. The SPACES_ROUTER_HOST is the domain name that's used to access the control plane instances. It's used by the load balancer or ingress to route requests.

export SPACES_ROUTER_HOST="proxy.upbound-127.0.0.1.nip.io"

important

Make sure to replace the placeholder text in SPACES_ROUTER_HOST and provide a real domain that you own.

Install the Spaces software

Install cert-manager

Install cert-manager.

kubectl apply -f https://github.com/cert-manager/cert-manager/releases/download/v1.19.3/cert-manager.yaml
kubectl wait deployment -n cert-manager cert-manager-webhook --for condition=Available=True --timeout=360s

Install AWS Load Balancer Controller

The AWS Load Balancer Controller provisions a Network Load Balancer when you expose the spaces-router Service with the annotations in the next step.

helm install aws-load-balancer-controller aws-load-balancer-controller --namespace kube-system \
  --repo https://aws.github.io/eks-charts \
  --set clusterName=${SPACES_CLUSTER_NAME} \
  --set serviceAccount.create=false \
  --set serviceAccount.name=aws-load-balancer-controller \
  --wait

Expose Spaces with LoadBalancer (recommended)

This guide exposes Spaces using a LoadBalancer Service on the spaces-router. No additional ingress or gateway components are required.

important

Use a Network Load Balancer (L4), not an Application Load Balancer (L7). Spaces uses long-lived connections for watch traffic that L7 load balancers may timeout.

Add the following to a values.yaml file (or use the same --set flags in the Helm install below). Replace proxy.example.com with your SPACES_ROUTER_HOST.

externalTLS:
  host: proxy.example.com  # Use your SPACES_ROUTER_HOST

router:
  proxy:
    service:
      type: LoadBalancer
      annotations:
        service.beta.kubernetes.io/aws-load-balancer-type: external
        service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing
        service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip

tip

To use Gateway API or Ingress instead of LoadBalancer, see Exposing Spaces externally.

Install Upbound Spaces software

Create an image pull secret so that the cluster can pull Upbound Spaces images.

kubectl create ns upbound-system
kubectl -n upbound-system create secret docker-registry upbound-pull-secret \
  --docker-server=https://xpkg.upbound.io \
  --docker-username="$(jq -r .accessId $SPACES_TOKEN_PATH)" \
  --docker-password="$(jq -r .token $SPACES_TOKEN_PATH)"

Log in with Helm to be able to pull chart images for the installation commands.

jq -r .token $SPACES_TOKEN_PATH | helm registry login xpkg.upbound.io -u $(jq -r .accessId $SPACES_TOKEN_PATH) --password-stdin

Install the Spaces software. Use the values.yaml file you created in the previous step (with externalTLS.host and router.proxy.service).

helm -n upbound-system upgrade --install spaces \
  oci://xpkg.upbound.io/spaces-artifacts/spaces \
  --version "${SPACES_VERSION}" \
  -f values.yaml \
  --set "account=${UPBOUND_ACCOUNT}" \
  --set "authentication.hubIdentities=true" \
  --set "authorization.hubRBAC=true" \
  --wait

Create a DNS record

important

Create a DNS record for the spaces-router load balancer before you create your first control plane.

Get the load balancer address for the spaces-router Service:

kubectl get svc -n upbound-system spaces-router \
  -o jsonpath='{.status.loadBalancer.ingress[0].hostname}'

If the preceding command doesn't return a load balancer address then your provider may not have allocated it yet. Once it's available, add a DNS record for the ROUTER_HOST to point to the given load balancer address. If it's an IPv4 address, add an A record. If it's a domain name, add a CNAME record.

Configure the up CLI

With your kubeconfig pointed at the Kubernetes cluster where you installed Upbound Spaces, create a new profile in the up CLI. This profile interacts with your Space:

up profile create --use ${SPACES_CLUSTER_NAME} --type=disconnected --organization ${UPBOUND_ACCOUNT}

Optionally, log in to your Upbound account using the new profile so you can use the Upbound Marketplace with this profile as well:

up login

Connect to your Space

Use up ctx to create a kubeconfig context pointed at your new Space:

up ctx disconnected/$(kubectl config current-context)

Create your first control plane

You can now create a control plane with the up CLI:

up ctp create ctp1

You can also create a control plane with kubectl:

cat <<EOF | kubectl apply -f -
apiVersion: spaces.upbound.io/v1beta1
kind: ControlPlane
metadata:
  name: ctp1
  namespace: default
spec:
  writeConnectionSecretToRef:
    name: kubeconfig-ctp1
    namespace: default
EOF

The first control plane you create in a Space takes around 5 minutes to get into a condition=READY state. Wait until it's ready using the following command:

kubectl wait controlplane ctp1 --for condition=Ready=True --timeout=360s

Connect to your control plane

Connect to your control plane with the up ctx command. With your kubeconfig still pointed at the Kubernetes cluster where you installed the Upbound Space, run the following:

up ctx ./default/ctp1

This command updates your current kubectl context. You're now connected to your control plane directly. Confirm this is the case by trying to list the CRDs in your control plane:

kubectl get crds

To disconnect from your control plane and switch back to your previous context:

up ctx -

tip

Learn how to use the up CLI to navigate around Upbound by reading the up ctx command reference.