9.1

Edit

Before you install

Before installing the WorkFusion software, read this section carefully and perform the required preparations.

Server preparation

Make sure the requirements given in the System requirements section are met for each server.

You may find on this page references to VDS acronym (Virtual Data Scientist). VDS is the former name for the AutoML component (Automatic Machine Learning). VDS means "AutoML" in all cases.

It's advised to use mounted HDD/SDD volume for the installation and keep it separate from the root volume where Operating System is installed.

caution

Workfusion SPA is designed to be installed as a non-privileged user. When first provisioning servers for WorkFusion SPA installation, make sure the system packages listed in Required Packages are installed.

If you skip this step, root permissions will be required to install these packages later.

Besides installation of required packages, root permissions will be required for execution of a small number of operations on the servers before actual installation. All these operations are documented in the Run preinstall manually section.

So, if during installation you will not have root permissions, request the root access or ask server administrators to execute everything from Run Preinstall Manually page.

If root access is available, you don't need to care about any preparations (aka preinstall steps). The installer will perform all of them automatically.

High-level installation flow

The information below will give you the big picture of what to expect during the installation of Workfusion SPA.

The entire Workfusion SPA product consists of a small number of components.

These components are logically grouped into seven roles that are distributed across seven servers as the default (the recommended architecture scheme):

  • DB
  • APP
  • APM
  • OCR
  • AutoML (AutoML-Standalone or AutoML-Cluster variations)
  • RPA
  • BI

Mind the following:

  • Each server role (APP, DB, APM, etc.) actually represents predefined set of software (workfusion-produced binaries and integrated third-party software).
  • Each server role is assumed to be installed on a separate server for best performance, although some custom combinations of components and their custom distribution across servers is supported, for example, for high availability scenarios.
  • Custom installation schemes can be performed individually per-customer on-demand.
  • The installation guide describes the default installation scheme. Components are distributed across servers as per scheme described in Product architecture section - seven servers with predefined set of software components.
  • Installation process is a defined set of actions which needs to be performed on each server.
  • DB server should be installed first. Other servers don't have dependencies from each other and can be installed in parallel or one-by-one in any order. Although the recommended sequence of installation:
    • Linux servers: DB > APP > APM > OCR > AutoML (Standalone or Cluster).
    • Windows servers: RPA, BI can be installed before or after installation of Linux servers or in parallel.
    • Windows servers: RPA server requires Administrative access on the system and permissions for RDP connection into the system. BI server requires Administrative access on the system for installation.
  • Installation of software components on each Linux server is performed using the same single installation package. It doesn't matter what server role you are provisioning (DB, APP, APM, OCR or AutoML), you'll use the same installation package and the same common config files for any of them.
  • Installation steps which need to be performed on each server are described further in documentation.

You don't need to perform the steps below right now. Just read them to get a big picture of what you are going to do. All the steps will be described in detail further in documentation in Preinstall steps and Installation steps sections. Installation process on each Linux server: DB, APP, APM, OCR, AutoML (vds-master, vds-agent, or vds-standalone) is divided into several parts:

  1. Uploading installation package to the server using any available method (scp, sftp, web, smb, nfs, rsynk, physical storage device, custom clients, for example, s3cmd, awscli, etc.).

  2. Unpacking installation package on the server.

  3. Filling up two config files (secrets.yml and config.yml) inside the folder where the package was unpacked.

  4. Executing the ./install.sh precheck <server_role> command in the directory where the package was unpacked. Available server roles: db, app, ocr, apm, vds-master, vds-slave, vds-standalone. This command only checks that the software can be installed on this particular server. Without passing this phase it is impossible to proceed with installation.

  5. On the APP server put your license.properties license file to the directory where you've extracted the installation package.

  6. Executing Preinstall steps. Preinstall steps require root access on the server. Two scenarios are available here:

    1. Fully automated preinstall steps performed by Workfusion-Installer with a single command: ./install.sh preinstall <server_role>. This is a recommended way. The command should be run with root user.
    2. Manual execution of Preinstall steps one-by-one. All preparation steps are documented in Run preinstall manually section for each particular Linux server.

    Each Linux server has common for all servers preinstall steps as well as its own set of preparation steps: DB, APP, APM, OCR, AutoML (vds-master, vds-slave, or vds-standalone).

  7. Preparation and putting the required TLS certificates into the certificates directory inside the folder where the package was unpacked as described in Using TLS certificates or automated generation of TLS certificates as also described in the Using TLS certificates section.

  8. Executing the ./install.sh install <server_role> command that performs the installation.

  9. Executing the ./install.sh check <server_role> command that checks the software is operational.

Password complexity verification

Since 9.0.0.8

The password complexity verification feature was implemented. We as a team encourage you to use stronger passwords for better security. The mandatory password complexity requirements will be listed below:

  • Password must be not shorter than 6 symbols and not longer than 20 symbols.
  • Password must contain at least one capital letter (A-Z).
  • Password must contain at least one number (0-9).
  • Password must contain at least one of the following cpecial symbols !*#%^@.

All passwords settings are configured in the secrets.yml file. If these rules aren't applicable, the password complexity verification can be turned off in the config.yml file with the help of the passwords check variable switched to false.

Domain name configuration

  • The steps below assume that you already have seven servers where Workfusion Software will be installed (5 Linux servers + 2 Windows servers).
  • Before performing any installation steps, configure DNS names for the servers as described in the DNS Names Configuration section.

Network availability between servers

  • The steps below assume that you already have seven servers where Workfusion Software will be installed (5 Linux servers + 2 Windows servers).
  • Check that the following network ports are opened between servers and for end users as described in Application ports section.

Installation packages

  • Download links for all required installation packages are provided by WorkFusion.
  • The full distribution of the Workfusion Software is a set of files (archives):
    • For DB, APP, APM, OCR, AutoML (standalone or cluster) Linux Servers:
      • workfusion-full-package-.tar.gz
    • For RPA Server (Windows):
      • RPAInstaller-.zip
      • Java SE Runtime Environment
    • For BI Server (Windows):
      • WorkFusionAnalyticsInstaller-.zip
      • WorkFusionAnalyticsWorkbooks-.zip

Linux installation package structure

The Main installation package workfusion-full-package-.tar.gz is used for provisioning all Linux servers has the following file structure.

AutoML installation types

AutoML supports two types of installation. Choose the one that best suits your needs.

  • AutoML Standalone:

    • AutoML Standalone: the simplest solution, all AutoML components (AutoML Master + NFS Share + one AutoML Agent) are installed and run** on a single server.
    • AutoML performance is limited by the server capacity.
    • May be converted to AutoML Cluster in future.
    • AutoML Standalone type of installation is shown on the Product architecture page in the high-level architecture diagram on top of the page.

  • AutoML Cluster:

    • AutoML Cluster: a combination of one AutoML Master + one or more AutoML Slaves.
    • Additional slaves may be configured and added as required at any point after the initial setup.
    • AutoML Cluster type of installation is shown on the Product architecture page in the AutoML Standalone vs AutoML Cluster diagram.

  • When choosing installation type AutoML Standalone or AutoML Cluster, ensure system requirements for your infrastructure are met as described in the System requirements section.

AutoML Standalone

  • AutoML Standalone is AutoML Master + NFS Share + One AutoML Agent installed on the same single server.
  • AutoML Standalone server can be installed following the instructions in Preinstall steps and Installation steps using vds-standalone as server_role.
  • AutoML Standalone may also be installed as sequential execution of all steps from Preinstall steps and Installation steps using vds-master as server_role firstly, and then using "vds-slave" as server_role on the same server.

AutoML Cluster

  • AutoML Cluster installation consists of a AutoML Master server + a number of separate AutoML Agent servers.
  • AutoML Cluster may also consist of AutoML Standalone + a number of separate AutoML Agent servers.
caution

Choose the NFS approach before proceeding.

AutoML Cluster uses NFS for sharing data between the master and the slaves. There are several ways of using NFS for the clustered AutoML setup.

  • NFS on AutoML Master: This is the default behavior when using clustered AutoML. No changes required.
  • NFS on a separate server (provided by customer): In case you have a dedicated NFS server for AutoML, it is necessary to set up and provide the shared location.
  • NFS on a separate server (configured via installer): NFS will be configured by the installer on a separate server. Follow the instructions in Preinstall steps and Installation steps using vds-nfs as server_role before installing AutoML Cluster.

Once you've chosen one of the options above, you should follow the requirements of that configuration according to the System requirements section. Re-configuring of an already installed AutoML NFS component is not supported.

Once NFS has been configured:

← PreviousNext →