Prerequisites for On-Premises Installation

Describes fabric node and user prerequisites for on-premises installation of the HPE Ezmeral Data Fabric.

Node Requirements for On-Premises Installation

Before deploying on-premises, you must provide the nodes that will host the on-premises fabric. Nodes that you want to include in the fabric must meet the following criteria:

One-Node Minimum Fabric
At least one node is required for an on-premises installation. You can provide as many additional nodes as you need.
Operating System
The following OS versions are supported:
  • RHEL 9.0, 8.8, 8.6, 8.5, 8.4, 8.3, 8.2, 8.1
  • Rocky 8.5, 8.4
  • Ubuntu 22.04, 20.04, 18.04
  • SLES 15 SP3, 15 SP2
  • OEL 8.4, 8.3, 8.2
Storage
For on-premises nodes, the HPE Ezmeral Data Fabric uses all the storage that is available to it. The platform does not impose limits based on the storage tier you select when creating the fabric. HPE recommends the following minimum number of nodes for each storage tier, but you do not need to follow this guideline precisely. The software does not currently check for a specific number of nodes or a specific amount of storage.
Storage Tier Recommended Minimum Number of Nodes
1 TB 3
10 TB 5
100 TB 7
1 PG 12
Fully Qualified Domain Names (FQDNs)
The nodes must be expressed as fully-qualified domain names (FQDNs). DO NOT specify hostnames as aliases or IP addresses.
Disk Space and Software Requirements
Nodes must meet the requirements in the following table. The Installer verifies the requirements prior to installation.
Table 1. Node Requirements
Component Requirements
CPU 64-bit x86.
CPU Cores Minimum of 16 per node
OS RHEL, Oracle Linux, Rocky, SLES, or Ubuntu.
Memory 32 GB minimum for nodes in production.
Disk Raw, unformatted drives and no partitions.
DNS Hostname, reaches all other nodes.
Users Common users across all nodes; passwordless ssh (optional).
Java Must run Java 11 or 17. Also, the Java and Java C versions must be the same on all nodes.
Python The default Python version must be set to Python 3 on all nodes.
Other NTP, Syslog, PAM.
Provide at least 10 GB of free disk space on the operating system partition. Provide 10 GB of free disk space in the /tmp directory and 128 GB of free disk space in the /opt directory. Services such as the ResourceManager and NodeManager use the /tmp directory. Files, such as logs and cores, use the /opt directory.
For data disks, the Installer requires a minimum disk size that is equal to the physical memory on the node. If a data disk does not meet the minimum disk size requirement, a verification error is generated.
Proxy Server Requirements
If nodes in the fabric use an HTTP proxy server, the nodes must also meet the following requirements:

  • The no_proxy environment variable must be set.

    Nodes in the fabric need to be able to communicate without the use of a proxy. If the https_proxy and http_proxy environment variable is set for nodes in the fabric, you must also set the no_proxy environment variable for the fabric admin user and the root user on each node. Configure the no_proxy environment variable to the IP range of the nodes or to the sub-domain that contains the nodes.

    In addition, you must follow this guideline from the Python documentation: "The no_proxy environment variable can be used to specify hosts which shouldn't be reached via proxy; if set, it should be a comma-separated list of hostname suffixes, optionally with :port appended, for example cern.ch,ncsa.uiuc.edu,some.host:8080."

    For cloud-based fabrics (Amazon EC2, Google Compute Engine (GCE), and Microsoft Azure), you must include this entry in the no-proxy configuration:
    169.254.169.254

  • The global proxy for package repositories must be set.

    The Installer creates repository files. However, the proxy setting is not configured for each repository. Therefore, configure global proxy settings on each node in the fabric.
    • On CentOS/RedHat, set global proxy settings in /etc/yum.conf.
    • On Ubuntu, set global proxy settings in /etc/apt/apt.conf.

User Requirements for On-Premises Installation

On-premises nodes must meet the following requirements for users:

  • For all users, the numeric user and group IDs (UID and GID) must be configured, and these values must match on all nodes in all fabrics.
  • The mapr user and root user must be configured to use bash. Other shells are not supported.
  • The user that initiates fabric creation for an on-premises deployment must be present and have the same user name and password on all fabric nodes.
  • Using an SSH key is not supported during on-premises fabric creation.