The MOSTLY AI Synthetic Data Platform comes packaged in two installation scripts. One is for installing the app server, and the second is for installing the AI server. This allows you to deploy multiple AI servers across multiple virtual machines. This guide will help you to prepare, run, and finalize the installation.

You can click on their names below to see their configuration parameters.

mostly-ai-<version>.sh, the app server installer
-h                Print this help message and exit

-b                Run install in batch mode (without manual
                  intervention)

-r                Completely remove a previous installation
                  before installation,
                  instead of updating the existing installation

-P <path>         Path prefix to install application to;
                  default: {{ root_path }} for
                  installation as a privileged and
                  {{ user_path }} for installation
                  as an unprivileged user

-D <path>         Central data storage path

-d <domain>       Domain the service will be reachable at;
                  default: -

-i <ip>           IP address to bind to;
                  default: 0.0.0.0 (all IPs)


-a <port>         Port the app web server listens at;
                  default: 8080


-k <port>         Port the user management KeyCloak listens at;
                  default: 8090


-c <port>         Port the coordinator listens at; default: 8081


-q <port>         Port the rabbitmq listens at; default: 5672


-K <path>         path to domain certificate for LDAPS


-L <path>         Location of the data files, where client can
                  store files for file-based catalog;
                  default:
                  <Central data storage path>/local-storage


-v                VM Management Type
                  [permanent, aws, gcp, azure];
                  default: permanent


-w                VM Management:AWS_Region;
                  Mandatory if VM Management Type='aws'



-z                VM Management:Azure_ResourceGroup;
                  Mandatory if VM Management Type='azure'


-g                VM Management:GCP_Application;
                  Mandatory if VM Management Type='gcp'


-p                VM Management:GCP_Project;
                  Mandatory if VM Management Type='gcp'


-o                VM Management:GCP_Zone;
                  Mandatory if VM Management Type='gcp'
mostly-agent-<version>.sh, the AI server installer
-h                Print this help message and exit

-b                Run install in batch mode
                  (without manual intervention)

-r                Completely remove a previous installation
                  before installation,
                  instead of updating the existing installation


-P <path>         Path prefix to install application to;
                  default: {{ root_path }} for
                  installation as a privileged and
                  {{ user_path }} for installation
                  as an unprivileged user


-D <path>         Central data storage path

-c <domain/ip>    Host Mostly-AI listens at;

-C <port>         Port the coordinator listens at; default: 8081


-q <domain/ip>    Host rabbitmq listens at;


-Q <port>         Port the rabbitmq listens at; default: 5672


-L <path>         Location of the data files,
                  where client can store files for file
                  based catalog; default:
                  <Central data storage path>/local-storage


-t <type>         [ALWAYS_ON/PRE_CREATED]
                  Specifies, if the installed agent is a
                  dynamic one (will be stopped and started by
                  coordinator), or it will be a static
                  always-on one; default: ALWAYS_ON

Installing the App server

Preparing the installation

  1. Before running the installer, you’ll first need to verify the minimum recommended partition sizes that MOSTLY AI requires to run successfully:

    • 30 GB for /var

    • 14-20 GB for /home

    • 14-20 GB for /tm


  2. You’ll also need to consider whether users require access to CSV files larger than 1 Gb. If this is the case, we recommend creating the following directory structure before installation:

    mostly-enterprise
    	┗ internal-data


Configuring the installtion parameters

Once you’ve verified the partition sizes and optionally created the directory structure for handling CSV files larger than 1 Gb, you can follow the below steps to gather the details for the installation parameters.

  1. The installer needs to know the paths to the application and internal data directory.
    Use the -P and -D parameters respectively to specify these.


  2. Next, you’ll need to tell the installer the domain name or IP address where users will access MOSTLY AI and Keycloak. The latter can then be accessed using [domain-name]/auth` (for example, mostlyai.mycompany.com/auth)

    Use the -d parameter to specify this domain name or IP address for the MOSTLY AI Synthetic Data Platform.

    Please keep in mind that these domain names only become reachable after the necessary network configurations (DNS servers, firewalls, etc.) have been performed.


  3. You’ll also need to tell the installer the server’s internal IP address (bind IP address). This allows the app to communicate with its internal components and the AI servers.

    Use the -i parameter to specify this internal IP address.

    An example command-line instruction, using the parameters discussed in the previous steps, would then look like this:

    bash mostly-ai-<version>.sh -b
    -d <domain name>
    -i <internal IP address>
    -P /home/ubuntu/mostly-enterprise/
    -D /home/ubuntu/mostly-enterprise/internal_data/


  4. Lastly, you’ll need to specify in what type of environment the app and AI servers will be running. You can choose between permanent for on-premises deployments, or aws, gcp, or azure for the respective cloud computing environments.

    Select your environment from the list below to see the required parameters and an example command-line instruction.

    On-premises
    -v                VM Management Type
                      [permanent, aws, gcp, azure];
                      default: permanent

    Example command-line instruction:

    bash mostly-ai-<version>.sh -b
    -d <domain name>
    -i <internal IP address>
    -P /home/ubuntu/mostly-enterprise/
    -D /home/ubuntu/mostly-enterprise/internal_data/
    -v permanent
    AWS
    -v                VM Management Type
                      [permanent, aws, gcp, azure];
                      default: permanent
    
    -w                VM Management:AWS_Region;
                      Mandatory if VM Management Type='aws'

    Example command-line instruction:

    bash mostly-ai-<version>.sh -b
    -d <domain name>
    -i <internal IP address>
    -P /home/ubuntu/mostly-enterprise/
    -D /home/ubuntu/mostly-enterprise/internal_data/
    -v aws
    -w <Valid AWS region name, e.g. eu-central-1>
    GCP
    -v                VM Management Type
                      [permanent, aws, gcp, azure];
                      default: permanent
    
    -g                VM Management:GCP_Application;
                      Mandatory if VM Management Type='gcp'
    
    
    -p                VM Management:GCP_Project;
                      Mandatory if VM Management Type='gcp'
    
    
    -o                VM Management:GCP_Zone;
                      Mandatory if VM Management Type='gcp'

    Example command-line instruction:

    bash mostly-generate-enterprise-1.4.5.sh -b
    -d <domain name>
    -i <internal IP address>
    -P /home/ubuntu/mostly-enterprise/
    -D /home/ubuntu/mostly-enterprise/internal_data/
    -v gcp
    -g <Valid GCP application name>
    -p <Valid GCP project ID>
    -o <Valid GCP zone name, e.g. europe-west1-b>
    Azure
    -v                VM Management Type
                      [permanent, aws, gcp, azure];
                      default: permanent
    
    -z                VM Management:Azure_ResourceGroup;
                      Mandatory if VM Management Type='azure'

    Example command-line instruction:

    bash mostly-ai-<version>.sh -b
    -d <domain name>
    -i <internal IP address>
    -P /home/ubuntu/mostly-enterprise/
    -D /home/ubuntu/mostly-enterprise/internal_data/
    -v azure
    -z <Valid Azure resource group>


Running the installer

You can now run the installer with all the needed installation parameters.
To execute the script as the current user, run:

bash mostly-ai-<version>.sh [parameters]

Or, use sudo for global installation:

sudo bash mostly-ai-<version>.sh [parameters]

You will see a few prompts during the installation. You can just press enter and skip them if you want to use the default settings. It won’t override any of the parameters you specified using the flags.

If you want to have a log file of the installation process, you can execute the following command:

bash -x mostly-generate-<version>.sh [parameters] > installation.log 2>&1

This command will produce a very large and detailed debug log, which you can use for in-depth analysis or audits.

Installing an AI server

To install an AI server, run mostly-agent-<version>.sh with the following parameters configured:

-D

The path to MOSTLY AI’s internal data directory.

-c

The IP address of the application server.

-q

The IP address of messaging server.

You can find an example command-line instruction here below:

bash mostly-agent-<version>.sh -b
-c <application server IP address>
-q <messaging server IP address>
-D /home/ubuntu/mostly-enterprise/internal_data/

Starting and Stopping MOSTLY AI

MOSTLY AI starts automatically once the installation has been completed. The installer will show the URL where users can access the web UI. You can control the application using the mostly-generate command. Just open a console and type:

  • sudo mostly-ai start to start the application

  • sudo mostly-ai stop to stop the application

  • sudo mostly-ai status to check the status

If the installation was not done via sudo you need to provide the full path to run mostly-ai, e.g.

[-P, application prefix path]/mostly-generate/bin/mostly-generate start

MOSTLY AI will start and show the URL for the web UI. An example looks like this:

Copy/paste this URL into your browser and you are ready to go!

License activation

Please follow the next steps to activate your license:

  1. In MOSTLY AI’s web UI, navigate to Settings > License and submit a request to activate your license by clicking on Create License Activation Request.

  2. A License Request modal appears. Enter your license code and click Create License Activation Request.

  3. Copy the activation request data and send it in an email to your MOSTLY AI Account manager.

  4. In return, you will receive an activation response from your Account manager.

  5. Click Activate License, insert the response in the activation response text field, and click Activate.

  6. After successful activation, the details of your current license will be displayed.

Updates

To perform an update, you’ll need to stop all Docker services using the sudo mostly-generate stop command and execute all steps in this installation guide. The installer will recognize and update the current installation of MOSTLY AI.