Deploy a Google Distributed Cloud cluster on OpenStack

This guide walks you through a sample deployment of Google Distributed Cloud on OpenStack virtual machines (VMs) with supported operating systems. The deployment uses a script to simplify installation of a hybrid cluster in OpenStack VMs. The guide also shows you one way to enable load balancing as a Service (LBaaS). You can use the OpenStack LBaaS and the Kubernetes OpenStack Cloud Provider in Google Distributed Cloud to expose the Kubernetes services outside of the OpenStack cluster.

Google Distributed Cloud doesn't provision the OpenStack VMs automatically, and provisioning the VMs is outside the scope of this guide. To learn the VM requirements and review an example deployment, see the Terraform example to create OpenStack VMs.

The guide consists of the following sections:

  1. Deploy Google Distributed Cloud

  2. Configure the OpenStack Cloud Provider for Kubernetes in Google Distributed Cloud to integrate with the Octavia load balancers

  3. Validate the OpenStack Cloud Provider for Kubernetes integration

This guide uses OpenStack Ussuri, but Google Distributed Cloud doesn't have a requirement for specific versions of OpenStack. The guide uses OpenStack VMs to provide you with a two-node Google Distributed Cloud proof of concept environment running on OpenStack. For information about creating a production environment with a high-availability control plane, see the Google Distributed Cloud documentation for production environment requirements.

Example deployment

This guide provides you with an example deployment of Google Distributed Cloud on OpenStack that integrates with OpenStack LBaaS. You must understand and adjust the commands and configuration values to suit your OpenStack environment. The following diagram shows the resulting deployment:

Google Distributed Cloud installed on OpenStack.


  • OpenStack Ussuri with LBaaS v2 deployed and functional
  • Service account for downloading the bmctl tool
  • Configure your OpenStack VMs and network as shown in the example deployment. To provision a similar setup in your OpenStack environment, you have the following options:
    1. Use this Terraform script to provision the resources automatically.
    2. Provision the resources manually.
  • The following OpenStack VMs must be ready and available through SSH:
Name IP address Purpose
abm-ws (private IP)
floating_ip (public IP)
Acts as the admin workstation: It's used to deploy Google Distributed Cloud to the other machines.
abm-cp1 GKE cluster control plane: This host runs the Kubernetes control plane and load balancer.
abm-w1 GKE cluster worker node: This host runs the Kubernetes workloads.

Deploy Google Distributed Cloud

This section shows you how to complete the following tasks:

  1. Install the tools you need on the abm-ws admin workstation VM.
  2. Configure the project ID and service account needed to securely complete the deployment
  3. Create a cluster configuration file
  4. Create a cluster

Install the tools you need

  1. Fetch the public floating IP address of the abm-ws VM:

    export OPENSTACK_IPS=$(openstack floating ip list --tags=abm_ws_floatingip -f json)
    export FLOATING_IP=$(jq -c '.[]."Floating IP Address"' <<< $OPENSTACK_IPS | tr -d '"')
  2. Ensure you can use SSH to connect securely into the abm-ws VM and sign in as a root user. The root user configured by the Terraform scripts is abm.

    ssh ubuntu@$FLOATING_IP
    sudo -u abm -i
  3. Verify that you can use SSH to connect to the other nodes:

    ssh abm@ 'echo SSH to $HOSTNAME succeeded'
    ssh abm@ 'echo SSH to $HOSTNAME succeeded'

    The expected responses for the preceding commands are:

    SSH to abm-cp1 succeeded
    SSH to abm-w1 succeeded
  4. Download the kubectl command line utility on the abm-ws VM.

    curl -LO "$(curl -s"
    chmod +x kubectl
    sudo mv kubectl /usr/local/sbin/
  5. Install Docker on the abm-ws VM:

    curl -fsSL -o
    sudo usermod -aG docker abm
    newgrp docker

Configure the Google Cloud project and service account

  1. Obtain Google Cloud CLI access credentials for your user account.

    These credentials are used with subsequent gcloud commands.

    gcloud auth login
  2. Make sure the Google Cloud CLI is configured to use the Google Cloud project in which you want your Google Distributed Cloud to be registered.

    gcloud config set project PROJECT_ID
  3. Set the Application Default Credentials (ADC) for your user account in the admin workstation. This will be used when the bmctl tool is used for cluster creation.

    gcloud auth application-default login
  4. Create the bm-gcr service account. You use this service account to authenticate from the Google Distributed Cloud cluster.

    gcloud iam service-accounts create bm-gcr
    gcloud iam service-accounts keys create bm-gcr.json \
  5. Enable the necessary APIs:

    gcloud services enable \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
  6. Give additional permissions to the bm-gcr service account. Adding the permissions means you don't need to create multiple service accounts for individual services.

    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \
    gcloud projects add-iam-policy-binding PROJECT_ID \
      --member="" \

Create a cluster configuration file

  1. Download the bmctl command line utility.

    mkdir baremetal && cd baremetal
    gsutil cp gs://anthos-baremetal-release/bmctl/1.29.200-gke.243/linux-amd64/bmctl .
    chmod a+x bmctl
    sudo mv bmctl /usr/local/sbin/
  2. Create an Google Distributed Cloud workspace for your cluster.

    bmctl create config -c CLUSTER_NAME
  3. Create the configuration file for your Google Distributed Cloud cluster.

    cat > bmctl-workspace/CLUSTER_NAME/CLUSTER_NAME.yaml << EOB
    gcrKeyPath: /home/abm/bm-gcr.json
    sshPrivateKeyPath: /home/abm/.ssh/id_rsa
    gkeConnectAgentServiceAccountKeyPath: /home/abm/bm-gcr.json
    gkeConnectRegisterServiceAccountKeyPath: /home/abm/bm-gcr.json
    cloudOperationsServiceAccountKeyPath: /home/abm/bm-gcr.json
    apiVersion: v1
    kind: Namespace
      name: openstack-cluster-ns
    kind: Cluster
      name: CLUSTER_NAME
      namespace: openstack-cluster-ns
      annotations: "true"
      type: hybrid
      anthosBareMetalVersion: 1.29.200-gke.243
        projectID: PROJECT_ID
          clusterName: CLUSTER_NAME
          - address:
        mode: manual
          controlPlaneLBPort: 443
        location: us-central1
        projectID: PROJECT_ID
          path: /mnt/localpv-disk
          storageClassName: node-disk
          numPVUnderSharedPath: 5
          path: /mnt/localpv-share
          storageClassName: standard
        loginUser: abm
    kind: NodePool
      name: node-pool-1
      namespace: openstack-cluster-ns
      clusterName: CLUSTER_NAME
      - address:

Create the cluster

  1. Create the cluster:

    bmctl create cluster -c CLUSTER_NAME

Running the bmctl command starts setting up a new hybrid cluster. This includes doing preflight checks on the nodes, creating the admin and user clusters and also registering the cluster with Google Cloud using Connect. The whole setup can take up to 15 minutes. You see the following output as the cluster is being created:

Please check the logs at bmctl-workspace/CLUSTER_NAME/log/create-cluster-20210926-020741/create-cluster.log
[2021-09-26 02:07:59+0000] Creating bootstrap cluster... ⠦ kind get kubeconfig --name bmctl > ~/.kube/config && k get pods --all-namespaces
[2021-09-26 02:07:59+0000] Creating bootstrap cluster... OK
[2021-09-26 02:10:48+0000] Installing dependency components... OK
[2021-09-26 02:13:42+0000] Waiting for preflight check job to finish... OK
[2021-09-26 02:15:22+0000] - Validation Category: machines and network
[2021-09-26 02:15:22+0000]  - [PASSED] gcp
[2021-09-26 02:15:22+0000]  - [PASSED] node-network
[2021-09-26 02:15:22+0000]  - [PASSED]
[2021-09-26 02:15:22+0000]  - [PASSED]
[2021-09-26 02:15:22+0000]  - [PASSED]
[2021-09-26 02:15:22+0000]  - [PASSED]
[2021-09-26 02:15:22+0000] Flushing logs... OK
[2021-09-26 02:15:23+0000] Applying resources for new cluster
[2021-09-26 02:15:24+0000] Waiting for cluster to become ready OK
[2021-09-26 02:25:04+0000] Writing kubeconfig file
[2021-09-26 02:25:04+0000] kubeconfig of created cluster is at bmctl-workspace/CLUSTER_NAME/CLUSTER_NAME-kubeconfig, please run
[2021-09-26 02:25:04+0000] kubectl --kubeconfig bmctl-workspace/CLUSTER_NAME/CLUSTER_NAME-kubeconfig get nodes
[2021-09-26 02:25:04+0000] to get cluster node status.
[2021-09-26 02:25:04+0000] Please restrict access to this file as it contains authentication credentials of your cluster.
[2021-09-26 02:25:04+0000] Waiting for node pools to become ready OK
[2021-09-26 02:25:24+0000] Moving admin cluster resources to the created admin cluster
[2021-09-26 02:25:53+0000] Flushing logs... OK
[2021-09-26 02:25:53+0000] Deleting bootstrap cluster...

Verify and interact with the cluster

You can find your cluster kubeconfig file on the abm-ws VM inside the bmctl-workspace directory. To verify your deployment, complete the following steps:

  1. Set the KUBECONFIG environment variable with the path to the cluster configuration file to run kubectl commands on the cluster:

    export KUBECONFIG=$HOME/bmctl-workspace/CLUSTER_NAME/CLUSTER_NAME-kubeconfig
    kubectl get nodes

    You should see the nodes of the cluster printed, similar to the following output:

    NAME      STATUS   ROLES                  AGE     VERSION
    abm-cp1   Ready    control-plane,master   5m24s   v1.20.5-gke.1301
    abm-w1    Ready    <none>                 2m17s   v1.20.5-gke.1301

Sign in to your cluster from Google Cloud console

To observe your workloads in the Google Cloud console, you must sign in to the cluster. For instructions and more information about signing in to your cluster, see Work with clusters from the Google Cloud console.

Clean up

You can clean up the cluster by issuing the following command in your admin workstation (abm-ws) VM:

export KUBECONFIG=$HOME/bmctl-workspace/CLUSTER_NAME/CLUSTER_NAME-kubeconfig
bmctl reset --cluster CLUSTER_NAME

What's next?