在Kubenetes集群上部署定制轻量级镜像部署Paddle Serving

参考文档：

https://github.com/PaddlePaddle/Serving/blob/v0.8.3/doc/Run_On_Kubernetes_CN.md
https://gitee.com/AI-Mart/Serving

步骤一

安装指令

yum install git 

步骤二

下载repo

git clone https://github.com/PaddlePaddle/Serving.git

步骤三

切换目录

cd Serving 

步骤四

查看能生成镜像的参数配置

bash tools/generate_runtime_docker.sh --help

生成特定镜像

bash tools/generate_runtime_docker.sh --env cuda10.1 --python 3.7 --serving 0.8.3 --paddle 2.2.2 --image_name paddle_serving_runtime:cuda10.1-py37-sv0.8.3-pd2.2.2

步骤五

登陆到镜像仓库

docker login -u xx -p xx xx-docker.pkg.coding.net

上传到特定仓库repo xx-docker.pkg.coding.net/xx/xx/xx

docker push xx-docker.pkg.coding.net/xx/xx/xx:cuda10.1-py37-sv0.8.3-pd2.2.2

步骤六

此时可以在dockerfile里面继承这个基础镜像做进一步的应用开发和部署

FROM xx-docker.pkg.coding.net/xx/xx/xx:cuda10.1-py37-sv0.8.3-pd2.2.2

COPY . /deploy
WORKDIR /deploy

# Install requirements
RUN pip config set global.index-url https://mirror.baidu.com/pypi/simple \
    && python3.7 -m pip install --upgrade setuptools \
    && python3.7 -m pip install --upgrade pip \
    && pip3.7 install -r requirements.txt \
    && python3.7 paddle_model.py


ENTRYPOINT python3.7 web_service.py