whl_deploy

whl_deploy is a powerful tool designed specifically for Apollo deployment, aiming to significantly simplify the setup of Apollo environments and the import/export of resources. By automating key steps, whl_deploy ensures an efficient, consistent, and seamless Apollo deployment experience.

✨ Key Features

whl_deploy provides three core functionalities, covering the essential aspects of the Apollo deployment lifecycle:

1. Automated Environment Setup: Installs and configures Docker, NVIDIA Container Toolkit, and optimizes the system, providing an ideal environment for Apollo.
2. Apollo Resource Import: Unified import of core Apollo components, including Docker images, source code, AI models, high-definition maps, and compiled caches, drastically reducing initial setup and startup times.
3. Resource Packaging & Distribution: Packages pre-configured or downloaded Apollo resources into reusable bundles, facilitating rapid deployment across multiple machines, especially useful in offline environments.

🚀 Quick Start

This section guides you on how to quickly install and use whl_deploy, focusing on the most common end-to-end deployment flow.

1. Installation

To ensure compatibility, please upgrade your setuptools and pip first:

# Highly Recommended: Upgrade setuptools and pip first
pip install setuptools -U
pip install --upgrade pip

# Install whl-deploy
pip install whl-deploy

2. End-to-End Deployment - Express Mode!

The following two commands are the essence of whl_deploy, covering the vast majority of Apollo deployment scenarios:

Step A: Prepare Host Environment

This command will automatically install Docker, the NVIDIA Container Toolkit, and perform necessary system configurations. This is the foundation for running Apollo.

whl-deploy setup all

Step B: Import Apollo Resources

After setting up the environment, you can import all pre-packaged core Apollo resources, such as source code, Docker images, AI models, etc., with a single command. Please ensure your source package is ready.

whl-deploy import all --package=source

---

3. More Granular Control

If you require more fine-grained control, whl_deploy also provides individual commands.

3.1. Host Environment Configuration - Step-by-Step

In some cases, you might want to configure the host environment step-by-step:

# Only install and configure Docker
whl-deploy setup docker

# Only install and configure NVIDIA Container Toolkit
whl-deploy setup nvidia_toolkit

3.2. Resource Import & Export - Category Management

whl_deploy allows you to manage various Apollo resources separately.

a. Source Code

• Import Source Code Package: Imports a zipped Apollo source code archive to a specified location.

  # China
  whl-deploy import source_code -i=https://gitee.com/daohu527/apollo-lite.git
  
  # github
  whl-deploy import source_code -i=https://github.com/wheelos/apollo-lite.git
  
  # local
  whl-deploy import source_code -i=apollo-lite-main.zip

• Export Source Code Package: Packages the source code from the current Apollo environment for reuse elsewhere.

  whl-deploy export source_code -o=apollo-lite-main.zip

b. Compiled Cache (Bazel Cache)

The Bazel compilation cache is crucial for accelerating the Apollo build process.

• Import Compiled Cache: Imports a pre-packaged Bazel cache.

  whl-deploy import cache -i=bazel_cache.tar.gz

• Export Compiled Cache: Exports the current Bazel cache for use across multiple machines or in future deployments.

  # only zip content
  whl-deploy export cache -o=bazel_cache.tar.gz

c. Docker Images

Manage Apollo's core container images.

• Import Docker Image: Imports a Docker image from a .tar file. This is useful for offline deployments or pre-loading images.

  whl-deploy import docker_image -i=whl_docker_image.tar

• Export Docker Image:
  • Export information about all currently recognized Apollo-related Docker images:

    whl-deploy export docker_image --info

  • Export a specific Docker image to a .tar file:

    whl-deploy export docker_image -i=nvidia/cuda:11.8.0-cudnn8-devel-ubuntu20.04 -o=cuda_image.tar

    • Note: Here, the -i parameter should specify the name and tag of the image to be exported. For example, to export an Apollo development image, use: whl-deploy export docker_image -i=apollo:dev-latest -o=apollo_dev.tar

d. High-Definition Maps (HD Maps) (TODO - To Be Implemented)

# all maps
whl-deploy import maps -i="map_data.tar.gz"

# san_mateo only
whl-deploy import maps -i="your_map_dir/san_mateo.tar.gz"

• whl-deploy export hd_maps -o=...

e. AI Models (TODO - To Be Implemented)

• whl-deploy import models -i=...
• whl-deploy export models -o=...

🤝 Contribution & Support

whl_deploy aims to simplify Apollo deployment. If you have any questions, suggestions, or wish to contribute code, feel free to submit an Issue or a Pull Request.