Browse Source

[to #43653669]feat: auto build docker images

auto build docker images
aone 任务: https://test.aone.alibaba-inc.com/jobs/1824567?buildId=143470479
修改任务分支
        Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/9566518

    * [to #43653669]feat: auto build docker images
master
mulin.lyh 3 years ago
parent
commit
01bb751425
8 changed files with 300 additions and 3 deletions
  1. +169
    -0
      .dev_scripts/build_image.sh
  2. +3
    -1
      .dev_scripts/ci_container_test.sh
  3. +3
    -2
      .dev_scripts/dockerci.sh
  4. +11
    -0
      .dockerignore
  5. +84
    -0
      docker/Dockerfile.ubuntu
  6. +15
    -0
      docker/rcfiles/conda.tuna
  7. +13
    -0
      docker/rcfiles/ubuntu20.04_sources.tuna
  8. +2
    -0
      requirements/runtime.txt

+ 169
- 0
.dev_scripts/build_image.sh View File

@@ -0,0 +1,169 @@
#!/bin/bash
# default values.
BASE_CPU_IMAGE=reg.docker.alibaba-inc.com/modelscope/ubuntu:20.04
BASE_GPU_IMAGE=reg.docker.alibaba-inc.com/modelscope/ubuntu:20.04-cuda11.3.0-cudnn8-devel
MODELSCOPE_REPO_ADDRESS=reg.docker.alibaba-inc.com/modelscope/modelscope
python_version=3.7.13
torch_version=1.11.0
cudatoolkit_version=11.3
tensorflow_version=1.15.5
modelscope_version=None
is_ci_test=False
is_dsw=False
is_cpu=False
run_ci_test=False
function usage(){
echo "usage: build.sh "
echo " --python=python_version set python version, default: $python_version"
echo " --torch=torch_version set pytorch version, fefault: $torch_version"
echo " --cudatoolkit=cudatoolkit_version set cudatoolkit version used for pytorch, default: $cudatoolkit_version"
echo " --tensorflow=tensorflow_version set tensorflow version, default: $tensorflow_version"
echo " --modelscope=modelscope_version set modelscope version, default: $modelscope_version"
echo " --test option for run test before push image, only push on ci test pass"
echo " --cpu option for build cpu version"
echo " --dsw option for build dsw version"
echo " --ci option for build ci version"
echo " --push option for push image to remote repo"
}
for i in "$@"; do
case $i in
--python=*)
python_version="${i#*=}"
shift
;;
--torch=*)
torch_version="${i#*=}"
shift # pytorch version
;;
--tensorflow=*)
tensorflow_version="${i#*=}"
shift # tensorflow version
;;
--cudatoolkit=*)
cudatoolkit_version="${i#*=}"
shift # cudatoolkit for pytorch
;;
--modelscope=*)
modelscope_version="${i#*=}"
shift # cudatoolkit for pytorch
;;
--test)
run_ci_test=True
shift # will run ci test
;;
--cpu)
is_cpu=True
shift # is cpu image
;;
--ci)
is_ci_test=True
shift # is ci, will not install modelscope
;;
--dsw)
is_dsw=True
shift # is dsw, will set dsw cache location
;;
--push)
is_push=True
shift # is dsw, will set dsw cache location
;;
--help)
usage
exit 0
;;
-*|--*)
echo "Unknown option $i"
usage
exit 1
;;
*)
;;
esac
done

if [ "$modelscope_version" == "None" ]; then
echo "ModelScope version must specify!"
exit 1
fi
if [ "$is_cpu" == "True" ]; then
export BASE_IMAGE=$BASE_CPU_IMAGE
base_tag=ubuntu20.04
export USE_GPU=False
else
export BASE_IMAGE=$BASE_GPU_IMAGE
base_tag=ubuntu20.04-cuda11.3.0
export USE_GPU=True
fi
if [[ $python_version == 3.7* ]]; then
base_tag=$base_tag-py37
elif [[ $python_version == z* ]]; then
base_tag=$base_tag-py38
elif [[ $python_version == z* ]]; then
base_tag=$base_tag-py39
else
echo "Unsupport python version: $python_version"
exit 1
fi

target_image_tag=$base_tag-torch$torch_version-tf$tensorflow_version
if [ "$is_ci_test" == "True" ]; then
target_image_tag=$target_image_tag-$modelscope_version-ci
else
target_image_tag=$target_image_tag-$modelscope_version-test
fi
export IMAGE_TO_BUILD=$MODELSCOPE_REPO_ADDRESS:$target_image_tag
export PYTHON_VERSION=$python_version
export TORCH_VERSION=$torch_version
export CUDATOOLKIT_VERSION=$cudatoolkit_version
export TENSORFLOW_VERSION=$tensorflow_version
echo -e "Building image with:\npython$python_version\npytorch$torch_version\ntensorflow:$tensorflow_version\ncudatoolkit:$cudatoolkit_version\ncpu:$is_cpu\nis_ci:$is_ci_test\nis_dsw:$is_dsw\n"
docker_file_content=`cat docker/Dockerfile.ubuntu`
if [ "$is_ci_test" != "True" ]; then
echo "Building ModelScope lib, will install ModelScope lib to image"
docker_file_content="${docker_file_content} \nRUN pip install --no-cache-dir modelscope==$modelscope_version -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html"
fi
echo "$is_dsw"
if [ "$is_dsw" == "False" ]; then
echo "Not DSW image"
else
echo "Building dsw image well need set ModelScope lib cache location."
docker_file_content="${docker_file_content} \nENV MODELSCOPE_CACHE=/mnt/workspace/.cache/modelscope"
fi
printf "$docker_file_content" > Dockerfile
docker build -t $IMAGE_TO_BUILD \
--build-arg USE_GPU \
--build-arg BASE_IMAGE \
--build-arg PYTHON_VERSION \
--build-arg TORCH_VERSION \
--build-arg CUDATOOLKIT_VERSION \
--build-arg TENSORFLOW_VERSION \
-f Dockerfile .

if [ $? -ne 0 ]; then
echo "Running docker build command error, please check the log!"
exit -1
fi
if [ "$run_ci_test" == "True" ]; then
echo "Running ci case."
export MODELSCOPE_CACHE=/home/mulin.lyh/model_scope_cache
export MODELSCOPE_HOME_CACHE=/home/mulin.lyh/ci_case_home # for credential
export IMAGE_NAME=$MODELSCOPE_REPO_ADDRESS
export IMAGE_VERSION=$target_image_tag
export MODELSCOPE_DOMAIN=www.modelscope.cn
export HUB_DATASET_ENDPOINT=http://www.modelscope.cn
export CI_TEST=True
export TEST_LEVEL=1
if [ "$is_ci_test" != "True" ]; then
echo "Testing for dsw image or MaaS-lib image"
export CI_COMMAND="python tests/run.py"
fi
bash .dev_scripts/dockerci.sh
if [ $? -ne 0 ]; then
echo "Running unittest failed, please check the log!"
exit -1
fi
fi
if [ "$is_push" == "True" ]; then
echo "Pushing image: $IMAGE_TO_BUILD"
docker push $IMAGE_TO_BUILD
fi

+ 3
- 1
.dev_scripts/ci_container_test.sh View File

@@ -16,5 +16,7 @@ if [ $? -ne 0 ]; then
echo "linter test failed, please run 'pre-commit run --all-files' to check"
exit -1
fi
# test with install
python setup.py install

PYTHONPATH=. python tests/run.py
python tests/run.py

+ 3
- 2
.dev_scripts/dockerci.sh View File

@@ -1,5 +1,4 @@
#!/bin/bash
IMAGE_NAME=reg.docker.alibaba-inc.com/dinger/modelscope
MODELSCOPE_CACHE_DIR_IN_CONTAINER=/modelscope_cache
CODE_DIR=$PWD
CODE_DIR_IN_CONTAINER=/Maas-lib
@@ -8,6 +7,7 @@ gpus='7 6 5 4 3 2 1 0'
cpu_sets='0-7 8-15 16-23 24-30 31-37 38-44 45-51 52-58'
cpu_sets_arr=($cpu_sets)
is_get_file_lock=false
CI_COMMAND=${CI_COMMAND:-'bash .dev_scripts/ci_container_test.sh'}
for gpu in $gpus
do
exec {lock_fd}>"/tmp/gpu$gpu" || exit 1
@@ -31,10 +31,11 @@ do
-e HUB_DATASET_ENDPOINT=$HUB_DATASET_ENDPOINT \
-e TEST_ACCESS_TOKEN_CITEST=$TEST_ACCESS_TOKEN_CITEST \
-e TEST_ACCESS_TOKEN_SDKDEV=$TEST_ACCESS_TOKEN_SDKDEV \
-e TEST_LEVEL=$TEST_LEVEL \
--workdir=$CODE_DIR_IN_CONTAINER \
--net host \
${IMAGE_NAME}:${IMAGE_VERSION} \
bash .dev_scripts/ci_container_test.sh
$CI_COMMAND
if [ $? -ne 0 ]; then
echo "Running test case failed, please check the log!"
exit -1


+ 11
- 0
.dockerignore View File

@@ -0,0 +1,11 @@
.gitignore
tests
data
.dev_scripts
.dockerignore
.git
.gitattributes
.pre-commit-config.yaml
.pre-commit-config_local.yaml
.readthedocs.yaml
Dockfile

+ 84
- 0
docker/Dockerfile.ubuntu View File

@@ -0,0 +1,84 @@
ARG BASE_IMAGE=reg.docker.alibaba-inc.com/modelscope/ubuntu:20.04-cuda11.3.0-cudnn8-devel
FROM $BASE_IMAGE
ARG DEBIAN_FRONTEND=noninteractive
ENV TZ=Asia/Shanghai
ENV CONDA_DIR /opt/conda
ENV PATH="${CONDA_DIR}/bin:${PATH}"
ENV arch=x86_64
SHELL ["/bin/bash", "-c"]
COPY docker/rcfiles /tmp/resources
RUN apt-get update && apt-get install -y --reinstall ca-certificates && \
cp /tmp/resources/ubuntu20.04_sources.tuna /etc/apt/sources.list && \
apt-get update && \
apt-get install -y locales wget git vim ffmpeg libsm6 tzdata language-pack-zh-hans ttf-wqy-microhei ttf-wqy-zenhei xfonts-wqy libxext6 build-essential ninja-build && \
wget https://packagecloud.io/github/git-lfs/packages/debian/bullseye/git-lfs_3.2.0_amd64.deb/download -O ./git-lfs_3.2.0_amd64.deb && \
dpkg -i ./git-lfs_3.2.0_amd64.deb && \
rm -f ./git-lfs_3.2.0_amd64.deb && \
locale-gen zh_CN && \
locale-gen zh_CN.utf8 && \
update-locale LANG=zh_CN.UTF-8 LC_ALL=zh_CN.UTF-8 LANGUAGE=zh_CN.UTF-8 && \
ln -fs /usr/share/zoneinfo/Asia/Shanghai /etc/localtime && \
dpkg-reconfigure --frontend noninteractive tzdata && \
apt-get clean && \
rm -rf /var/lib/apt/lists/*

ENV LANG=zh_CN.UTF-8 LANGUAGE=zh_CN.UTF-8 LC_ALL=zh_CN.UTF-8

#install and config python
ARG PYTHON_VERSION=3.7.13
RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-${arch}.sh -O ./miniconda.sh && \
/bin/bash miniconda.sh -b -p /opt/conda && \
rm -f miniconda.sh && \
ln -s /opt/conda/etc/profile.d/conda.sh /etc/profile.d/conda.sh && \
echo ". /opt/conda/etc/profile.d/conda.sh" >> ~/.bashrc && \
cp /tmp/resources/conda.tuna ~/.condarc && \
source /root/.bashrc && \
conda install --yes python==${PYTHON_VERSION} && \
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

ARG USE_GPU=True

# install pytorch
ARG TORCH_VERSION=1.12.0
ARG CUDATOOLKIT_VERSION=11.3
RUN if [ "$USE_GPU" = "True" ] ; then \
conda install --yes pytorch==$TORCH_VERSION torchvision torchaudio cudatoolkit=$CUDATOOLKIT_VERSION -c pytorch && conda clean --yes --all; \
else \
conda install pytorch==$TORCH_VERSION torchvision torchaudio cpuonly -c pytorch; \
fi

# install tensorflow
ARG TENSORFLOW_VERSION=1.15.5
RUN if [ "$USE_GPU" = "True" ] ; then \
pip install --no-cache-dir --use-deprecated=legacy-resolver tensorflow==$TENSORFLOW_VERSION -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html; \
else \
pip install --no-cache-dir tensorflow==$TENSORFLOW_VERSION; \
fi

RUN if [ "$USE_GPU" = "True" ] ; then \
CUDA_HOME=/usr/local/cuda TORCH_CUDA_ARCH_LIST="5.0 5.2 6.0 6.1 7.0 7.5 8.0 8.6" MMCV_WITH_OPS=1 MAX_JOBS=8 FORCE_CUDA=1 pip install --no-cache-dir mmcv-full && pip cache purge; \
else \
MMCV_WITH_OPS=1 MAX_JOBS=8 pip install --no-cache-dir mmcv-full && pip cache purge; \
fi

# install modelscope
COPY requirements /var/modelscope
RUN pip install --no-cache-dir --upgrade pip && \
pip install --no-cache-dir -r /var/modelscope/runtime.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html && \
pip install --no-cache-dir -r /var/modelscope/audio.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html && \
pip install --no-cache-dir -r /var/modelscope/cv.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html && \
pip install --no-cache-dir -r /var/modelscope/multi-modal.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html && \
pip install --no-cache-dir -r /var/modelscope/nlp.txt -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html && \
pip cache purge

# default shell bash
ENV SHELL=/bin/bash

# install special package
RUN pip install --no-cache-dir mmcls>=0.21.0 mmdet>=2.25.0 decord>=0.6.0 numpy==1.18.5 datasets==2.1.0

RUN if [ "$USE_GPU" = "True" ] ; then \
pip install --no-cache-dir dgl-cu113 dglgo -f https://data.dgl.ai/wheels/repo.html; \
else \
pip install --no-cache-dir dgl dglgo -f https://data.dgl.ai/wheels/repo.html; \
fi

+ 15
- 0
docker/rcfiles/conda.tuna View File

@@ -0,0 +1,15 @@
channels:
- defaults
show_channel_urls: true
default_channels:
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
custom_channels:
conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
pytorch-lts: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud

+ 13
- 0
docker/rcfiles/ubuntu20.04_sources.tuna View File

@@ -0,0 +1,13 @@
# 默认注释了源码镜像以提高 apt update 速度,如有需要可自行取消注释
deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal main restricted universe multiverse
# deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal main restricted universe multiverse
deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-updates main restricted universe multiverse
# deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-updates main restricted universe multiverse
deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-backports main restricted universe multiverse
# deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-backports main restricted universe multiverse
deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-security main restricted universe multiverse
# deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-security main restricted universe multiverse

# 预发布软件源,不建议启用
# deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-proposed main restricted universe multiverse
# deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ focal-proposed main restricted universe multiverse

+ 2
- 0
requirements/runtime.txt View File

@@ -8,6 +8,8 @@ numpy
opencv-python
oss2
Pillow>=6.2.0
# for pyarrow 9.0.0 event_loop core dump
pyarrow>=6.0.0,!=9.0.0
pyyaml
requests
scipy


Loading…
Cancel
Save