Skip to content

installation_process.md

Latest commit

 

History

History
180 lines (121 loc) · 3.68 KB

installation_process.md

File metadata and controls

180 lines (121 loc) · 3.68 KB

Pulse Installation Guide

This document describes the full installation process for deploying Pulse in a Dataiku environment.

It assumes familiarity with Dataiku administration concepts such as plugins, code environments, Code Studios, and API keys.

1. Required Permissions

Platform Admin Access

To install Pulse, you must be a Dataiku Platform Admin on the primary (core) Dataiku instance.

This is required because Pulse installation involves:

  • Installing a Dataiku plugin
  • Creating a Code Environment
  • Creating a Code Studio template
  • Creating and managing Dataiku projects
  • Running administrative macros

In practice, full admin access on the core instance is required.

2. Multi-Instance API Access

Pulse can collect metadata and usage data from multiple Dataiku instances.

For each instance you want to connect:

  • You must be a full admin
    OR
  • An admin must generate an API key with sufficient permissions

API Key Capabilities

API keys must be able to:

  • Read metadata
  • Read audit and usage logs

These API keys are configured during Pulse setup.

3. Plugin Installation

  1. Log in to Dataiku as an administrative user

  2. Navigate to Waffle → Plugins

  3. Install the plugin from Git:

    https://github.com/dataiku/dss-plugin-dataiku-pulse.git

  4. Build the plugin code environment

    • No containers are required

4. Plugin Configuration

After installation, open the Plugin Settings page.

Primary Parameter Set

Create a single parameter set named:

primary

⚠️ The name must be lowercase.

Populate the following sections:

GitHub Repository

  • Repository URL
  • Branch (typically main)

Dashboard Configuration

  • Dashboard Project Key 
    DATAIKU_PULSE_DASHBOARD
  • Dashboard Host URL (hostname or IP:port)
  • Dashboard Host API Key (admin-level)
  • Blob Storage Connection (AWS, Azure, or GCS)

Worker Nodes

  • Worker Project Key 
    DATAIKU_PULSE_WORKER
  • One entry per Dataiku instance:
    • Hostname or IP:port
    • Admin-level API key
  • User to own and run scenarios
  • Optional settings:
    • Ignore certificates
    • Project data parallelization
    • Core count

Optional Per-Host Parameter Sets

You may create additional parameter sets for host-specific overrides:

  • Custom user
  • Certificate behavior
  • Parallelism / core settings

5. Code Studio Setup

  1. Navigate to Code Studios

  2. Create a template named:

    dataiku_pulse_dashboard

    ⚠️ This name is required

  3. Configure the template to:

    • Run on Kubernetes
    • Use the Dataiku Pulse (Streamlit Custom) block
    • Disable user permissions

  4. Build the template

6. Project Initialization

  1. Create the Pulse Dashboard Project

  2. Navigate to Macros

  3. Filter on:

    Dataiku Pulse: Initialize

  4. Run:

    • Initialize Dashboard
    • Initialize Workers

  5. Navigate to Code → Code Studios

  6. Publish the application as a Web Application

    • Auto-start enabled

⚠️ The dashboard may initially appear empty while the first collection cycle runs.

7. Blob Storage & DuckDB

Pulse stores all collected data in external blob storage and loads final tables into DuckDB.

  • AWS and Azure work out of the box using standard credentials
  • GCS requires additional IAM configuration

See docs/gcs-auth.md for GCS-specific setup and validation.

8. Installation Complete

Once the initial collection cycle completes, the Pulse dashboard will begin displaying insights and usage metrics.

At this point:

  • Core installation is complete
  • Additional customization and extensions can be added as needed