Deploy Union.ai on CKS

Documentation Index

Fetch the complete documentation index at: https://docs.coreweave.com/llms.txt

Use this file to discover all available pages before exploring further.

​

Prerequisites

• A CKS cluster with kubectl access configured.
• The helm CLI installed. For installation instructions, see Install Helm.
• The uctl CLI installed.
• The flyte CLI installed.
• A Union.ai organization. Contact Union.ai to create one if you don’t have one.

​

Configure CoreWeave AI Object Storage

​

Create a bucket

​

Generate access credentials

​

Create an access policy

{
  "policy": {
    "version": "v1alpha1",
    "name": "union-ai-bucket-access",
    "statements": [
      {
        "name": "allow-union-s3-access",
        "effect": "Allow",
        "actions": ["s3:*"],
        "resources": [
          "[BUCKET-NAME]",
          "[BUCKET-NAME]/*"
        ],
        "principals": ["*"]
      }
    ]
  }
}

​

Configure the Union CLI

admin:
  endpoint: dns:///[ORG-NAME].cloud-staging.union.ai
  insecure: false
  authType: Pkce

uctl get project

​

Provision data plane resources

export KUBECONFIG=[PATH-TO-KUBECONFIG]

uctl selfserve provision-dataplane-resources \
  --clusterName [CLUSTER-NAME] \
  --provider metal

​

Create the Helm values file

host: "[ORG-NAME].cloud-staging.union.ai"
clusterName: "[CLUSTER-NAME]"
orgName: "[ORG-NAME]"
provider: metal

storage:
  provider: custom
  bucketName: "[BUCKET-NAME]"
  fastRegistrationBucketName: "[BUCKET-NAME]"
  custom:
    type: stow
    container: "[BUCKET-NAME]"
    stow:
      kind: s3
      config:
        region: "[AVAILABILITY-ZONE]"
        auth_type: accesskey
        access_key_id: "[ACCESS-KEY-ID]"
        secret_key: "[SECRET-ACCESS-KEY]"
        endpoint: https://cwobject.com
        disable_ssl: false
        disable_force_path_style: "true"

executor:
  extraEnvVars:
    - name: FLYTE_AWS_S3_ADDRESSING_STYLE
      value: "virtual"
    - name: FLYTE_AWS_ENDPOINT
      value: "https://cwobject.com"
    - name: AWS_ACCESS_KEY_ID
      value: "[ACCESS-KEY-ID]"
    - name: AWS_SECRET_ACCESS_KEY
      value: "[SECRET-ACCESS-KEY]"
    - name: AWS_DEFAULT_REGION
      value: "[AVAILABILITY-ZONE]"

config:
  k8s:
    plugins:
      k8s:
        default-env-vars:
          - FLYTE_AWS_ENDPOINT: "https://[BUCKET-NAME].cwobject.com"
          - FLYTE_AWS_S3_ADDRESSING_STYLE: "virtual"
          - AWS_ACCESS_KEY_ID: "[ACCESS-KEY-ID]"
          - AWS_SECRET_ACCESS_KEY: "[SECRET-ACCESS-KEY]"
          - AWS_DEFAULT_REGION: "[AVAILABILITY-ZONE]"

operator:
  enableTunnelService: true

secrets:
  admin:
    create: true
    clientId: [CLIENT-ID]
    clientSecret: [CLIENT-SECRET]

fluentbit:
  enabled: true
  env:
    - name: AWS_ACCESS_KEY_ID
      value: [ACCESS-KEY-ID]
    - name: AWS_SECRET_ACCESS_KEY
      value: [SECRET-ACCESS-KEY]

​

Deploy the Union data plane

1. Clone the Union Helm charts repository and check out the latest stable release: Replace [RELEASE-TAG] with the latest stable release tag (for example, dataplane-2026.3.3). Check the Union Helm charts repository for available tags.
   git clone https://github.com/unionai/helm-charts.git
   cd helm-charts
   git checkout [RELEASE-TAG]
   
   After checkout, you have the CRD and dataplane charts for the release you install.
2. Install the Custom Resource Definitions (CRDs):
   helm upgrade --install unionai-dataplane-crds charts/dataplane-crds
   
3. Build the chart dependencies and install the data plane: Replace [PATH-TO-VALUES-FILE] with the path to the Helm values file you customized in the previous section.
   cd charts/dataplane
   helm dependency build
   helm upgrade --install unionai-dataplane . \
     --namespace union --create-namespace \
     --values [PATH-TO-VALUES-FILE] \
     --timeout 10m
   
4. Verify the deployment:
   kubectl get pods -n union
   
   When the deployment succeeds, all pods show a Running status, including union-operator-proxy, union-operator-buildkit, flytepropeller, and executor.

​

Register the cluster

1. Clone the Union cloud repository and create a branch for your organization: Replace [CLOUD-REPO-URL] with your organization’s cloud repository URL. Replace [ORG-NAME] with your organization name.
   git clone [CLOUD-REPO-URL]
   cd cloud
   git checkout -b [ORG-NAME]
   
2. Generate tenant configurations:
   make generate_tenant_configs
   
3. Commit and push the generated configurations:
   git add .
   git commit -m "Generate tenant configs for [ORG-NAME]"
   git push origin [ORG-NAME]
   
4. Trigger a Buildkite build for your organization’s staging pipeline. In Buildkite, start new build (not Rebuild) for the pipeline. That starts a pipeline run that picks up the branch you pushed.
5. Verify cluster registration:
   uctl get cluster
   
   The output is similar to the following:
   NAME               ORG          STATE          HEALTH
   union-coreweave    my-org       STATE_ENABLED  HEALTHY
   

​

Create the eager API key

uctl create apikey --keyName EAGER_API_KEY --org [ORG-NAME]

​

Test a workflow

1. Create a Flyte CLI configuration file at the path .flyte/config.yaml in your project directory: Replace [ORG-NAME] and [PROJECT-NAME] with your organization and project identifiers.
   admin:
     endpoint: dns:///[ORG-NAME].cloud-staging.union.ai
   image:
     builder: remote
   task:
     domain: development
     org: [ORG-NAME]
     project: [PROJECT-NAME]
   
2. Run a sample workflow:
   flyte run --image ghcr.io/flyteorg/flyte:py3.13-v2.0.2 \
     hello_world.py main --n 5
   
   If the remote image builder isn’t enabled for your organization, use the --image flag with a pre-built container image as in the preceding flyte run example.
3. Check the run status: Replace [RUN-NAME] with the workflow run identifier.
   flyte get run [RUN-NAME]
   
   Look for ACTION_PHASE_SUCCEEDED in the output.

​

Troubleshooting

​

Additional resources

• Union.ai documentation
• Flyte documentation
• CoreWeave AI Object Storage
• Create a CKS cluster