amjames / qhub

QHub deployment tool

Information	Links
Project
Community
CI

Table of contents

• Table of contents
• QHub HPC
• QHub
  • ☁️ Cloud Providers
• 💻 Installation
  • Pre-requisites
  • Install QHub
• 🏷️ Usage
• ❓ Questions?
• 📖 Code of Conduct
• ⚙️ Installing the Development version of QHub
• 🙌 Contributions
• Ongoing Support
• License

Automated data science platform. From JupyterHub to Cloud environments with Dask Gateway.

QHub is an open source tool that enables users to build and maintain cost-effective and scalable compute/data science platforms on HPC or on Kubernetes with minimal DevOps experience.

This repository details the QHub (Kubernetes) version.

Not sure what to choose? Check out our Setup Initialization page.

QHub HPC

Version of QHub based on OpenHPC.

NOTE: The tool is currently under development. Curious? Check out the Qhub HPC repository.

QHub

The Kubernetes version of QHub is built using Terraform, Helm, and GitHub Actions. Terraform handles the build, change, and versioning of the infrastructure. Helm helps to define, install, and manage Kubernetes. GitHub Actions is used to automatically create commits when the configuration file (qhub-config.yaml) is rendered, as well as to kick off the deployment action.

QHub aims to abstract all these complexities for its users. Hence, it is not necessary to know any of the above mentioned technologies to have your project successfully deployed.

TLDR: If you know GitHub and feel comfortable generating and using API keys, you should have all it takes to deploy and maintain your system without the need for a dedicated DevOps team. No need to learn Kubernetes, Terraform, or Helm.

☁️ Cloud Providers

QHub offers out-of-the-box support for Digital Ocean, Amazon AWS, GCP, and Microsoft Azure.

For more details, check out the release blog post.

💻 Installation

Pre-requisites

• QHub is supported by macOS and Linux operating systems (Windows is NOT currently supported).
• Compatible with Python 3.7+. New to Python? We recommend using Anaconda.
• Adoption of virtual environments (conda, pipenv or venv) is also encouraged.

Install QHub

To install QHub type the following commands in your command line:

• Install using conda:

  conda install -c conda-forge qhub

• Install using pip:

  pip install qhub

Once finished, you can check QHub's version (and additional CLI args) by typing:

qhub --help

If successful, the CLI output will be similar to the following:

usage: qhub [-h] [-v] {deploy,destroy,render,init,validate} ...

QHub command line

positional arguments:
  {deploy,destroy,render,init,validate}
                        QHub

optional arguments:
  -h, --help            show this help message and exit
  -v, --version         QHub version

🏷️ Usage

QHub requires the setting of environment variables to automate the deployments fully. For details on how to obtain those variables, check the installation guide in the docs.

Once all the necessary credentials are gathered and set as UNIX environment variables, QHub can be deployed in under 20 minutes using:

qhub init   ... # generates initial config file and optionally automates deployment steps
qhub deploy ... # creates and configures the cloud infrastructure

❓ Questions?

Have a look at our FAQ to see if your query has been answered.

We separate the queries for QHub into:

• GitHub Discussions used to raise discussions about a subject, such as: "What is the recommended way to do X with QHub?"

• Issues for queries, bug reporting, feature requests,documentation, etc.

We work around the clock to make QHub better, but sometimes your query might take a while to get a reply. We apologise in advance and ask you to please, be patient 🙏.

📖 Code of Conduct

To guarantee a welcoming and friendly community, we require contributors to follow our Code of Conduct.

⚙️ Installing the Development version of QHub

To install the latest developer version (unstable) use:

pip install git+https://github.com/Quansight/qhub.git@dev

🙌 Contributions

Thinking about contributing? Check out our Contribution Guidelines.

Ongoing Support

The v0.4.0 release introduced many changes that will irrevocably break your deployment if you attempt an inplace upgrade; for details see our RELEASE notes. In order to focus on the future direction of the project, we have decided as a team that we will provide limited support for older versions. Any new user is encouraged to use v0.4.0 or greater.

If you're using an older version of QHub and would like professional support, please reach out to Quansight.

License

QHub is BSD3 licensed.