Installer FAQ

Review frequently asked questions about the Installer.

General

What is the difference between the Installer and the Quick Installer?

The Installer is a robust, user-friendly replacement for the Quick Installer. You can use the Installer to install a cluster with HPE Ezmeral Data Fabric services and ecosystem components. You can also use the Installer to update an existing cluster with additional nodes, HPE Ezmeral Data Fabric services, and ecosystem components. However, the Installer does not install the client.

Which versions of software can I install?

You can use the Installer to install core releases 6.x and 7.x. Earlier core releases have reached their "End of Maintenance" status. For more information about the supported versions of HPE Ezmeral Data Fabric core software, see Core Support and Lifecycle Status. To understand which Installer versions can be used with older core versions, see Selecting an Installer Version to Use.

Can I use the Installer to upgrade my cluster?

Yes. The Installer can be used to upgrade a cluster that was installed using the Installer or an Installer Stanza. See Upgrading Core With the Installer for information about how to upgrade with the Installer.

If the cluster was manually installed, you can install the Installer and enable it to be used with subsequent installations or upgrades by following the steps in Using probe and import to Generate the Installer Database.

Can I use the Installer to install a patch?

Yes. See Applying a Patch Using the Installer.

Does the Installer support adding a "compute-only" node?
A compute-only node is a node that is capable of performing computational tasks but is not expected to perform long-term data storage. The Installer does not explicitly support adding a compute-only node to a cluster. However, you can effectively work around the issue by adding a node that has the file system and sufficient associated disk space. The disk space must be equal to or greater than the amount of physical memory on the node.

Preparing to Install

What are the Installer requirements?

See the Prerequisites.

What information should I have before I start?
The mapr-setup.sh script requests the following information:
  • The fully-qualified domain name for each host and the port number that other nodes in the cluster can use to connect to the Installer node.
  • A user for the cluster admin user account. If the user account doesn't exist, the mapr-setup.sh script prompts for the UID, GID, group name, and password so that it can create the account.

The Installer requests the following information:

  • The EEP that you want to install on a 5.2.x or later cluster
  • The services that you want to install on the cluster
  • Hostnames of the nodes that you want to include in the cluster (specify fully-qualified domain names as described in Connectivity)
  • Credentials for the root user or a user with sudo privileges on each node in the cluster
What are the node requirements?

See the Prerequisites.

What are my options if I don't want to use an Internet repository?

See Using a Local, Shared Repository With the Installer.

Using the Installer

Which license edition applies to my installation?
As of release 5.1, licenses are categorized by new editions and modules that further define the features supported by an edition.

See the following table for descriptions of the license options. For more information about licensing, see HPE Ezmeral Data Fabric Software Licensing.

License Edition Description
Community Edition An unlimited, free, community-supported edition with one free NFS Gateway. This edition includes Hadoop, HPE Ezmeral Data Fabric Database, and HPE Ezmeral Data Fabric Streams. However, real-time global replication of HPE Ezmeral Data Fabric Database tables orHPE Ezmeral Data Fabric Streams is not included.
Enterprise Edition Edition that enables enterprise-class features such as high availability, multi-tenancy, and disaster recovery. Each of the following modules for the Enterprise Edition unlocks a portion of the total platform capabilities:
Analytics
Enables enterprise-class features for analytic use cases, such as highly-available NFS and support for services like YARN and MapReduce.
Database
Enables enterprise-class features for operational NoSQL database, with HPE Ezmeral Data Fabric Database JSON and binary tables, and real-time global database replication.
Streams
Enables enterprise-class features for publish/subscribe event streaming, with HPE Ezmeral Data Fabric Streams and real-time global stream replication.

For more information about editions, see the ALA link at the bottom of this page.

What happened to the M3, M5, M7, or Enterprise Edition licenses?
With the release of 5.1 and Streams, the licensing model has been simplified, allowing more choice in which specific features are licensed on a cluster.

See the following table to understand how the new licenses correspond to the legacy license editions that you are familiar with. For more information about licensing, see Product Licensing. For more information about editions, see the ALA link at the bottom of this page.

Legacy Edition New Edition & Module(s)
M3 or Community Edition Community Edition. Starting in 5.1, the Converged Community Edition includes Streams.
M5 or Enterprise Edition Enterprise Edition with Hadoop Module
M7 or Enterprise Database Edition Enterprise Edition with Hadoop and Database modules
What expressions can I use to specify multiple nodes?

You can enter the following types of expressions to specify nodes:

  • [0-99] => Expands hostnames to 0, 1, 2, ...99. The second delimiter allows one or more digits.
  • [00-99] => Expands hostnames to 00,01,02,...99. Allows two or more digits of the same length
  • [a-z] or [A-Z] => Expands hostnames to a,b,c,...z or A,B,C,...Z

To group hosts based on racks for performance or reliability, append ":" followed by the rack name to each expression.

Examples:

  • host1, host2, host3
  • host[A-Z][0-99] => hostA0, hostA1, hostA2 ,...hostZ99
  • host[000-333] => host000, host001, host002, ... host333
  • host[0-3],otherhost[00-05] => host0, host1, host2,...host3 and otherhost00,...otherhost05
  • host[0-5]:rack1,host[6-10]:rack2 => host1, host2, host3,...host5 on rack1. host6, host7, host8,...host10 on rack2
How do I change the service layout?

The Installer uses groups to organize nodes and services. A group is a set of services that you can run on one or more nodes. A service can only be assigned to one group.

On the Configure Service Layout page, you can use the Advanced Configuration option to drag and drop services between existing groups to specify where the services are to be installed. You can also create new groups or change the list of nodes assigned to a group.
NOTE Some services can only be assigned to one node.
Can I install a single-master service, such as Hive, on multiple nodes?
When you use the Installer, single-master services are added to the default MASTER group. By design, only one node can be assigned to the MASTER group.

To install a single-master service on more than one node:

  1. On the Configure Service Layout page, create a new group.
  2. Assign multiple nodes to the new group.
  3. Drag the single-master service to the new group.
Can I install a secure cluster with the Installer?

Yes. Installer versions 1.10 and later support security by default.

Are there limitations to what you can do when you update an existing cluster?

See the restrictions documented in Using the Incremental Install Function.

How do I uninstall the Installer?

See Uninstalling Software Using the Installer Uninstall Button. If the installer node is part of the cluster, the Installer packages can remain on the installer node after the cluster is uninstalled.

If you have uninstalled the cluster, you can also run one of the following commands to uninstall the Installer packages from the installer node:

  • On CentOS / Red Hat:

    yum remove 'mapr-installer*'
  • On Ubuntu:
    apt-get remove 'mapr-installer*'
  • On SLES
    zypper remove 'mapr-installer*'

Troubleshooting

What can I do if I need to rerun mapr-setup.sh with a new repository URL and the script still points to the old URL?

The /opt/mapr/installer/data/properties.json file stores information such as the user ID of the cluster administrator, the user ID of the Installer, the OS type, Internet access information, and the repository URLs for Core and the ecosystem components. Once a repository URL has been stored in properties.json, the Installer assumes that the URL will not change. Rerunning the setup script does not update the URL. Even upgrading the installer packages does not update the repository URL in properties.json. To pass a new repository value into properties.json, you have two options:

Option 1

You can remove the installer files and rerun the setup script:
mapr-setup.sh remove

Using the remove command removes properties.json, the installer database, and the installer packages, but not the setup script. After the files are removed, you can rerun the setup script to specify the new repository URL. For more information about options you can use with mapr-setup.sh, see Using mapr-setup.sh.

Option 2

If you need to retain the installer database and the cluster state information (for example, because you need to do an upgrade), you can:
  1. Edit the properties.json file manually to change the repo_core_url and the repo_eco_url entries to the correct values.
  2. Restart the Installer:
    systemctl restart mapr-installer
What can I do if drop-down menus aren't working?
Try any or all of the following to correct the problem:
  • Refresh the browser page.
  • Clear the browser cache.
  • Close and restart the browser or browser tab.
Why does the Installer URL not work?

Check that the URL you are trying to access is external. For example, if you install on a cluster that is in the cloud, the URL that the Installer lists may not work if it is an internal URL. Try accessing the external URL that is associated with the internal URL.

Why doesn't the Installer list the ecosystem component that I want to install?

The Installer Definitions package contains the versions and services that you can install. Once you update the Installer Definitions, you can install ecosystem components that were made available after you first configured the Installer. See Updating the Installer.

On the Verify Nodes page, how can I get more information about a warning or error?

Hold your cursor over the warning or error in the right pane to see more information about the specific warning or error condition.

Why are the nodes listed on the Verify Nodes page different from those that I chose to install on?

If you abort an installation and then install on a different set of nodes, you must use the Verify Nodes page to manually remove nodes that were part of the aborted installation but are no longer part of the current installation.

I can't log in to Hue. What credentials should I use to log into Hue for the first time?
Log in with the cluster administrator username that you configured while running mapr-setup.sh and the password mapr.
What should I do if rerunning mapr-setup.sh generates errors because the properties.json file has incorrect information?
See Troubleshooting Repository URL Errors.