APRICOTLab

APRICOTLab: Advanced Platform for Reproducible Infrastructure in the Cloud via Open Tools for JupyterLab.

---

🌟 Introduction

APRICOTLab is an open-source JupyterLab extension that enables users to deploy, manage, and interact with multi-cloud infrastructures using a step-by-step graphical wizard and built-in IPython magics. It empowers reproducible research and experimentation in cloud-based environments using the familiar interface of Jupyter notebooks.

🔍 Features

• Deploy cloud infrastructures using a visual wizard or programmatically

• Support for multiple cloud providers like EGI, OpenStack or EC2

• Customizable infrastructure templates

• Integrated IPython magic commands for managing and interact with infrastructures directly from notebooks

• Seamless command execution, file upload/download, and monitoring

• Create reproducible computational experiments

🧬 Experiment Replication Methodology

APRICOTLab enables reproducible experiments that require complex infrastructure by:

• Providing a set of predefined, configurable infrastructures that can be redeployed by other researchers using the same configuration.

• Allowing command execution on deployed infrastructures directly from the notebook, making it easy to document and install additional software during an experiment.

• Leveraging the Jupyter environment for full documentation of experiments with text, live code, and visual outputs.

⚙️ Requirements

A Jupyter environment (compatible with JupyterLab ≥ 4.0.0).

The Infrastructure Manager (IM) client must be installed and configured, an open-source virtual infrastructure provisioning tool for multi-Clouds.

✨ IPython Magics for Infrastructure Management

The extension provides a set of custom IPython magic commands for interacting with deployed infrastructures:

To use the magics, load them in your notebook:

%reload_ext apricot_magics

You can also configure Jupyter to load them automatically.

Line magics

• %apricot_token [refresh_token]: Generates a new access token using the provided refresh_token. If no refresh_token is supplied, a previously saved token will be used instead.

• %apricot_ls: Lists all deployed infrastructures.

• %apricot_log <infra_id>: Shows the deployment logs of the specified infrastructure.

• %apricot_info <infra_id>: Returns the specifications of the given infrastructure.

• %apricot_upload <infra_id> <local_paths> <dest_path>: Uploads local files to the specified infrastructure.

• %apricot_download <infra_id> <remote_paths> <local_dest>: Downloads files from the infrastructure to the local system.

• destroy <infra_id>: Destroys the specified infrastructure.

Line and cell magic

• %%apricot (or %apricot): Supports subcommands such as:

• exec <infra_id> <command>: Executes a shell command in the specified infrastructure.

• list: Alias for %apricot_ls.

• destroy <infra_id>: Destroys the specified infrastructure.

🐳 Docker

A Dockerfile is provided for easy setup:

docker build -t apricotlab .
docker run --publish 8888:8888 apricotlab

This will launch a JupyterLab instance with APRICOT pre-installed.

🛠 Development Installation

Note: Node.js is required for building the extension. Note: The jlpm command is JupyterLab's pinned version of yarn that is installed with JupyterLab.

# Clone the repository
git clone https://github.com/grycap/apricotlab.git
cd apricotlab

# Install the Python package in development mode
pip install -e .

# Link the extension to JupyterLab
jupyter labextension develop . --overwrite

# Build the extension
jlpm build

You can run the extension in watch mode:

# Terminal 1: Watch the source directory, automatically rebuilding when needed
jlpm watch

# Terminal 2: Run JupyterLab
jupyter lab

To ensure source maps for easier debugging:

jupyter lab build --minimize=False

🧹 Uninstall (Development)

pip uninstall apricot

You should also remove the symlink created with jupyter labextension develop. Run:

jupyter labextension list

...to find the labextensions folder and manually remove the apricot symlink.

📦 Packaging the Extension

For packaging and releasing, see RELEASE.md.

🛡️ License

Distributed under the BSD 3-Clause License.