building_with_docker.md

Developing XGBoost Serving with Docker

The recommended approach to build XGBoost Serving from source is to use Docker. The XGBoost Serving Docker development images encapsulate all the dependencies you need to build your own version of XGBoost Serving.

Note: Currently we only support building binaries that run on Linux.

Installing Docker

General installation instructions are on the Docker site.

Clone the source

After installing Docker, we need to get the source we want to build from. We will use Git to clone the xgboost branch of XGBoost Serving:

git clone https://github.com/hzy001/xgboost-serving.git
cd xgboost-serving

Pulling a development image

For a development environment where you can build XGBoost Serving, you can try:

docker pull hzy001/xgboost-serving:latest-devel

Building the model server

Run the following command to generate the tensorflow_model_server executable:

./tools/run_in_docker.sh bazel build -c opt //tensorflow_serving/model_servers:tensorflow_model_server

Binaries are placed in the bazel-bin directory, and can be run using a command like:

./tools/run_in_docker.sh ./bazel-bin/tensorflow_serving/model_servers/tensorflow_model_server

Building the Python API PIP package

Run the following command to generate the build_pip_package executable:

./tools/run_in_docker.sh bazel build -c opt //tensorflow_serving/tools/pip_package:build_pip_package

Binaries are placed in the bazel-bin directory, and you can run the following command to generate the Python API PIP package:

./tools/run_in_docker.sh ./bazel-bin/tensorflow_serving/tools/pip_package/build_pip_package $(pwd)

If the command finishes successfully, you will see the tensorflow_serving_api-2.0.0-py2.py3-none-any.whl package in the current directory and you can install this package to use the tensorflow serving apis.