Prepare the hardware, operating system, and local repositories

Prepare each cluster node for a Data Fabric installation on a hardened operating system by configuring the hardware, installing a supported operating system, and setting up local package repositories for offline installation.

Before you install Data Fabric on a hardened operating system, every node in the cluster must have:
  • Healthy, correctly configured hardware
  • A supported operating system installed and on the network
  • Access to local package repositories, so you can install without Internet access

Complete these tasks on every node unless noted otherwise.

Configure the hardware environment

  1. Check the hardware. Use your server's management interface to confirm that there are no hardware faults and that the firmware is current.
  2. Configure storage. Use the server's storage or RAID configuration utility to remove any existing configuration. Then create separate logical volumes for the operating system and for Data Fabric:
    Volume Purpose Recommendation
    OS volume Operating system installation Use a redundant RAID level, such as RAID 1.
    Data Fabric volume Data Fabric storage Choose a RAID level based on your performance and capacity needs.
  3. Check the configuration. Save the storage configuration, reboot the node, and confirm that the logical volumes are visible.

For detailed steps, see your hardware vendor's documentation.

Install the operating system

  1. Install a supported OS. Install a hardened operating system that the Data Fabric release supports. Include the base system, legacy compatibility, Python 3, and server application packages or modules that Data Fabric needs.
  2. Configure networking. Give each node a static IP address and a fully qualified domain name (FQDN). Set the default gateway, DNS servers, and search domain. If you use network bonding or VLANs, configure them now, and make sure the switch configuration matches.
  3. Partition the OS disk. Install the OS only on the OS volume. Leave the Data Fabric volume unpartitioned. Make sure there is enough space for /, /tmp, /opt, and swap, as described in the Data Fabric installation prerequisites.
  4. Set up user accounts. Create the mapr user on every node. Use the same UID, GID, and password on all nodes. The root password must also be the same on all nodes.
  5. Check the installation. After the node restarts, confirm that it is reachable on the network, that SSH works, and that the OS version, IP address, and hostname are correct.
  6. Configure DNS. Make sure forward and reverse DNS records exist for every node.

Create the local repositories

If the nodes cannot reach the Internet, create local repositories so the package manager can install the OS and Data Fabric packages.
  1. Create an OS repository. Copy the OS installation media (for example, the ISO image) to the node, or to a shared location. Then add it as a repository using your OS's package management tools.
  2. Create a Data Fabric repository.

    1. Install the repository creation tool for your OS, such as createrepo_c.

    2. Create a directory for the repository, for example /var/www/html/repo/mapr.

    3. Copy the Data Fabric Core and Ecosystem Pack package archives into that directory and extract them:
            tar -xzvf mapr-v<core-version>.rpm.tgz
      tar -xzvf mapr-eep-v<eep-version>.rpm.tgz
  3. Add the directory as a local repository in your package manager.
  4. Check the repositories. List the configured repositories and confirm that both the OS and Data Fabric repositories are available.
TIP
As an alternative to building repositories on each node, you can use the Air-Gap Repository Container to host the Data Fabric packages.