Fabric Deployment Using a Seed Node

Describes how to install the platform using a seed node and the Create Fabric interface.

The first fabric you deploy – whether it is in the cloud or on-premises – must be deployed using a seed node. The seed node creates a lightweight, temporary fabric. The installation sequence uses this temporary fabric to display the interface for fabric creation.

The seed node is needed only for creation of the first fabric. Any additional fabric must be created from your cloud-based or on-premises fabric. The seed node can only be used to create fabrics. Creating volumes, buckets, or other resources from the seed node is not supported. For information about creating additional fabrics, see Creating a Fabric.

Seed Node Deployment Process

To deploy the HPE Ezmeral Data Fabric in your environment, you run a script that starts a Docker container. The Docker container emulates the behavior of a Data Fabric node. This emulated node is the seed node. The seed node enables access to the Create Fabric interface used to create fabrics.

Once the seed node is created and started, the system prints a URL that you use to access the Create Fabric interface. The interface enables you to complete the steps required to deploy a fabric in the cloud or on-premises. After the fabric is created, you can spawn additional fabrics from any installed fabric:

Prerequisites for Fabric Deployment

Before you deploy a fabric, review all of the following requirements:

Seed Node Prerequisites

Verify that the node you plan to use as the seed node meets the following prerequisites:
Prerequisite Notes
Users To install using a seed node, you must be root or a user that can run sudo commands without being prompted for a password.
OS The seed node has been tested on the following operating systems, but it can work on other operating systems and in other environments where Docker containers are hosted:
  • Mac OS 10
  • Ubuntu 20.04
Connectivity The seed node that you use to host the Docker container can be a server or laptop that supports the following, but must have connectivity to all the subnets and the VPN for the cloud provider:
  • Docker*
  • Bash
  • SSH
*On Ubuntu, RHEL, and Fedora hosts where Docker is not present, Docker is installed automatically. Installation of Docker can take 5-7 minutes. On Mac hosts where Docker is not present, you must install Docker before proceeding with the seed-node deployment.
CPU 64-bit x86 with a minimum of 16 cores per node.
Memory On the seed node, enough memory must be allocated to Docker to enable the container to come up and run. Docker must be installed and have the following memory allocated to it:
  • Mac: At least 32 GB
  • Other platforms: At least 8 GB
This is a general recommendation. Sometimes the container can run with less memory. On a seed node with many containers competing for memory, this recommendation might not be sufficient.
Disk Space If you use a Linux server for the seed node, allocate at least 50 GB of disk space to run the container.
Proxy If the seed node is behind a proxy, update the proxy configuration in /etc/environment using the following commands. You must do this before attempting to create a fabric:
export http_proxy=<http://myproxy.net>
export https_proxy=<http://myproxy.net>
export HTTP_PROXY=<http://myproxy.net>
export HTTPS_PROXY=<http://myproxy.net>
To specify the proxy information in a file when you run the setup script, see Help for datafabric_container_setup.sh.

On-Premises Deployment Prerequisites

See Prerequisites for On-Premises Installation.

Cloud Provider Prerequisites

You must have sufficient permissions to perform tasks in the cloud environment you are using. Hewlett Packard Enterprise recommends the following minimum permissions for installers:
Cloud Provider Minimum Permissions
AWS AmazonEBSCSIDriverPolicy and AmazonEC2FullAccess
Azure Contributor role
GCP Editor role
Before starting the deployment, gather the information that you will need to fill out the fabric-creation form for your cloud service provider. You can record the necessary information in the Planning Worksheet for Cloud Deployments.

Docker Image Prerequisites

The Docker image that must be used for HPE Ezmeral Data Fabric deployments is:

maprtech/edf-seed-container:7.9.0_9.3.1_edf

Users can also specify the maprtech/edf-seed-container:latest tag, which is always the latest version of the image.
IMPORTANT
Only an image that has the _edf suffix in the image tag can be used to install the consumption-based, software-as-a-service platform.
The Docker image name includes the Data Fabric core and ecosystem pack version information:

You can download the image from the Docker hub at:

https://hub.docker.com/r/maprtech/edf-seed-container/tags/

Or you can use the optional docker pull command as described in the following steps.

After you verify that your system and environment meet the listed prerequisites, complete the following steps to deploy the HPE Ezmeral Data Fabric.

Run the Script to Bring up the Container Image on the Seed Node

Use the following steps to bring up the container image and launch the Create Fabric interface:
  1. Sign in to the seed node host as root or a user that can run sudo commands without being prompted for a password.
  2. Download thedatafabric_container_setup.sh script from GitHub. For example, download the script in its raw form by using the following wget command:
    wget https://raw.githubusercontent.com/mapr-demos/edf-seednode-790-getting-started/main/datafabric_container_setup.sh
  3. Optional: Use a docker pull command to pre-download a copy of the image from https://hub.docker.com/r/maprtech/edf-seed-container/tags. Using docker pull requires sudo privileges:
    docker pull maprtech/edf-seed-container:latest
    IMPORTANT
    Note these considerations:
    • Pre-downloading is optional, but it makes the script run faster and prevents download issues when you run the script. The script checks to see if the image is already present on your system. If the image is present, the script uses the image. If it is not present, the script tries to download it.
    • If you want to use the -i <image> option, you must specify:
      maprtech/edf-seed-container:7.9.0_9.3.1_edf
  4. Modify the script so it is executable:
    chmod +x datafabric_container_setup.sh
  5. Before running the script, review the following considerations:
    • Running the datafabric_container_setup.sh script requires sudo privileges.
    • The script can take 5-10 minutes to run the first time you run it. Unless you pre-downloaded the image, the script downloads the latest Docker image from the Docker repository. For a list of available tags, see https://hub.docker.com/r/maprtech/edf-seed-container/tags.
    • The script supports the -i option for specifying an image other than the latest image. The script also supports a -p option that must be used if the seed node is a cloud instance. To view the command line help for the script, see Help for datafabric_container_setup.sh.
  6. Run the script to deploy the container for the Data Fabric image:
    ./datafabric_container_setup.sh
    The script downloads the latest image. When the Docker image is running, you see the following output:
    ./datafabric_container_setup.sh
                RAM NEEDED : AVAILABLE
                DOCKER STATUS : RUNNING
                PORTS NEEDED : AVAILABLE
                PROCEEDING FORWARD WITH DEPLOYING SEED NODE
Please enter the local sudo password for root

latest: Pulling from maprtech/edf-seed-container
Digest: sha256:b863f487de7eaa809b66f923aaec297ae1cab4fdb9441950fbe5c68328235a7a
Status: Downloaded newer image for maprtech/edf-seed-container:latest
docker.io/maprtech/edf-seed-container:latest
Developer Sandbox Container 8935a11fb7d6 is running..
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up

Client has been configured with the docker container.

Please click on the link https://<hostname>:8443/app/dfui to deploy data fabric
For user documentation, see https://docs.ezmeral.hpe.com/datafabric/home/installation/installation_main.html
    If the services do not come up, see Troubleshooting Seed Node Installation.
  7. Navigate to the link specified in the Docker output message in step 6. The Data Fabric UI displays the Create your first fabric form.

Create the Fabric

Bringing up the seed node automatically displays the fabric-creation interface. To create your first permanent fabric, complete the following steps.

  1. On the Create your first fabric form, fill in the configuration parameters for the type of fabric you want to create. For more information, see the following topics:

  2. Click Create. The Fabric details dialog box is displayed. For example:

  3. To monitor the progress of fabric creation, check the status bar in the Fabric details dialog box, or click See details. Fabric creation can take 20 minutes or more.

    While the installation is in progress, you can Cancel the installation at any time. If you click Cancel, a confirmation box asks for your credentials based on the provider type (AWS, GCP, Azure, or on-premises). Once you provide your credentials and click Continue, you cannot resume the current installation. However, you can use the Re-initiate button later to trigger a new installation request.

    If fabric creation is successful, the Data Fabric UI displays a message with the endpoint link for the new fabric. For example:

  4. Using a browser, navigate to the endpoint link, and sign in to the Data Fabric UI for the newly created fabric using user name admin and password p@ssw0rd. These credentials are the default user name and password for Keycloak.
    IMPORTANT
    Hewlett Packard Enterprise recommends that you change the default Keycloak password immediately after installation. See Changing the Keycloak admin Password.


  5. If fabric creation is successful, use the following steps to activate and register the fabric.
    NOTE
    After you have successfully created a cloud-based or on-premises fabric, you can kill the container that hosts the edf-installer.hpe.com fabric. For example, at the Docker command line type:
    % docker kill <seed-node-container-ID>
  6. If an error occurs during fabric creation, see Troubleshoot Fabric Creation later on this page.

Activate and Register the Fabric

After the first fabric is created, perform these steps:
  1. Add the activation key for the fabric. See Adding an Activation Key.
  2. Register the fabric. See Registering a Fabric.
  3. Set the billing model. See Setting the Billing Model.
  4. Configure single sign-on. See SSO Using Keycloak.

Troubleshoot Fabric Creation

Fabric creation can fail for various reasons. You might see a message like this:

To view the log information, click See details. For example:

Review the log information to determine if the failure is correctable. If you can resolve the failure condition, you can retry creating the fabric.
The Reinitiate button allows you to clean up failed fabric resources and trigger a fresh fabric request. To reinitiate fabric creation:
  1. Click Reinitiate. The Reinitiate fabric form is displayed.
  2. Fill in the required fields in the form.
  3. Click Create. The Fabric details dialog box is displayed.
  4. Monitor the progress of fabric creation as described in Create the Fabric.
To delete a failed fabric without reinitiating fabric creation, click Delete on the Fabric details dialog box.

If you are not able to resolve the issue that caused fabric creation to fail, contact HPE Support.