App Jupyter

JupyterHub Installation

Explanation (Core Concepts)

What is JupyterHub Installation?

JupyterHub Installation in xDP deploys a multi-user, production-ready Jupyter notebook environment directly onto one of your xDP compute clusters. It gives data science and analytics teams a secure, scalable, pre-configured workspace for interactive exploration and machine learning — without anyone hand-assembling Spark, kernels, and storage connectivity. You install it once from the xDP Apps catalog, and every notebook session inherits the cluster's compute and data access.

Under the hood, JupyterHub is a long-running app installed via a Helm chart on the target compute cluster. The control plane orchestrates the deployment and stores the rendered configuration; the cluster runs the hub, the proxy, and an optional observability connector. JupyterHub does not wire up data sources at install time — it configures Spark, catalog, and storage access at notebook start time, so each user session gets its own pod with the right configuration.

Key Concepts

• Apps Catalog — The Apps page where curated tools (JupyterHub, Spark, Trino, Airflow, NiFi) install onto a compute cluster with a guided wizard.

• Compute Cluster — The target cluster (selected in the top-right switcher) where JupyterHub deploys and where notebook workloads run.

• Helm Chart Version — The packaged version you install; constrained by your cluster's release version.

• Internal vs External PostgreSQL — JupyterHub needs PostgreSQL for hub metadata (users, server state, sessions). Let xDP deploy and manage one automatically (internal), or point it at your own managed instance (external, recommended for production).

• xObserve Connector — An optional xo-jupyter-connector that streams Jupyter and Spark telemetry into xObserve; requires xObserve on the linked xCentral.

• Deployment Configuration — The Helm values (editable YAML) controlling culling, base URL, named servers, ECR credential refresh, and more.

Capabilities

• Guided provisioning — A five-step wizard takes you from version selection to a running JupyterHub server.

• Flexible metadata storage — Zero-config internal PostgreSQL for development, or an enterprise-managed external database for production. The choice is locked once an installation exists.

• Optional observability — Enable the xo-jupyter-connector to feed metrics into xObserve; skippable and disabled until xObserve exists on xCentral.

• YAML-level customization with validation — Edit Helm values via Form or raw YAML; you must Validate before continuing.

• Notebook-time data access — Data and catalog access resolves when a notebook starts, not at install.

Tutorial (Getting Started)

This tutorial installs JupyterHub using the internal PostgreSQL database (recommended external is shown too).

Prerequisites

• Apache Spark already installed on the target cluster.

• An active compute cluster selected in the switcher, and permission to install applications on it.

• For an external database (optional): a reachable PostgreSQL instance, its JDBC URL, and credentials, with egress allowed from the cluster.

• To enable observability (optional): xObserve installed on the linked xCentral.

Your First Workflow

1. Confirm the correct cluster is selected, open Apps, locate the Jupyter Hub card, and click Install.

2. On Step 1 — Release Version, review the App Release Version and Cluster Release Version, choose the Helm Chart Version, and click Next.

3. On Step 2 — Configure PostgreSQL, choose the metadata database:

• External PostgreSQL (Recommended) — provide the Database URL (JDBC), Username, and Password of your managed instance.

• Internal PostgreSQL — xDP provisions and manages a dedicated database automatically. This guide selects Internal.

Click Next.

4. On Step 3 — xObserve Integration (optional), leave the connector off if xObserve is not installed, then click Next (or Skip This Step).

5. On Step 4 — Configure Application, review the generated Deployment Configuration. Click Validate (defaults are sufficient for a standard install), then Next.

6. On Step 5 — Complete, xDP applies the configuration and deploys JupyterHub. Click Finish to return to Apps.

7. The Jupyter Hub card now shows Installed, with xObserve, Edit, and Uninstall controls. The hub becomes usable once deployment finishes.

How-to Guides

Use an external PostgreSQL database

1. On Configure PostgreSQL, select the External PostgreSQL card.

2. Enter the Database URL (jdbc:postgresql://hostname:5432/database), Username, and Password.

3. Ensure the cluster has network egress to the database host, then click Next.

Customize the deployment with the YAML editor

1. On Configure Application, use the YAML view (toggle between Form and YAML).

2. For example, tune idle culling:

3. Click Validate and resolve any TO_BE_UPDATED placeholders (or Skip & Validate to bypass warnings), then Next.

Enable xObserve observability for JupyterHub

1. Ensure xObserve is installed on the linked xCentral; until then the toggle stays disabled.

2. Edit the app and, on xObserve Integration, enable the connector, then complete the wizard.

Reference

Configuration options

Services

Best Practices

• Install Spark first — Jupyter's Install stays disabled until Spark is present on the cluster.

• Use an external database in production — managed PostgreSQL with backup/HA/DR, decided before the first install.

• Validate before you continue and clear TO_BE_UPDATED placeholders.

• Version-control your YAML to review, roll back, and reproduce deployments.

• Install xObserve first if you want observability — the connector stays disabled until it exists on xCentral.

• Tune idle culling so abandoned notebooks don't hold cluster CPU/memory.