FATE 1.7.2编译arm64镜像示例，apple silicon与鲲鹏处理器均已测试

[kanppa]

本示例目前仅适用FATE 1.7.2基于eggroll部署的版本，并且没有nn模块。于apple m1 上与 鲲鹏cpu上测试通过

1. 拉取FATE代码

git clone https://github.com/FederatedAI/FATE.git -b v1.7.2 --recurse-submodules --depth=1

# clone compile scripts and move to FATE directory
git clone https://github.com/kanppa/FATE-ARM64.git

cp -r FATE-ARM64/build-arm FATE

cd FATE

git submodule update

2.编译arm64镜像

cd build-arm/docker-build

bash build_cluster_docker.sh all

docker images

编译之后会显示以下镜像

REPOSITORY               TAG             IMAGE ID       CREATED        SIZE
federatedai/eggroll      1.7.2-release   532f4f92fcfc   10 days ago    6GB
federatedai/fateboard    1.7.2-release   1b35e811aae1   10 days ago    347MB
federatedai/python       1.7.2-release   0a4a4a2f3e01   10 days ago    5.83GB
federatedai/base-image   1.7.2-release   85f1cc1092e4   10 days ago    5.56GB

3. docker-compose部署

git clone https://github.com/FederatedAI/KubeFATE.git -b v1.7.2

3.1 替换模版

需要 https://github.com/kanppa/FATE-ARM64 对项目部分文件进行替换

docker-compose-eggroll.yml
替换 
KubeFATE/docker-deploy/training_template/docker-compose-eggroll.yml

3.2 for apple silicon

如果是用的apple silicon，由于部分指令不通用
需要用 项目中的以下文件对脚本进行替换

mac-os/generate_config.sh
mac-os/docker_deploy.sh

并且安装 gtar

brew install gnu-tar

3.3 部署

参考 https://github.com/FederatedAI/KubeFATE/blob/v1.7.2/docker-deploy/README_zh.md

4. 测试

docker exec -it confs-${party_id}_python_1 /bin/bash

进入容器之后

cd fate

#client也会自动安装
pip install -e python/fate_test

# configure FATE-Pipeline server setting
pipeline init --port 9380 --ip 127.0.0.1
# configure FATE-Flow Commandline server setting
flow init --port 9380 --ip 127.0.0.1


fate_test config edit

cd /data/projects/fate
fate_test suite -i examples/dsl/v2/

如有疑问可以参考
https://github.com/FederatedAI/FATE/blob/v1.7.2/doc/tutorial/fate_test_tutorial.md

[dylan-fan]

推荐，