Adding Dockerfile for PT1.5 (#71)

TusharKanekiDey · Tushar Dey · web-flow · commit 99655b262fa0 · 2020-04-27T19:40:25.000-07:00
* Pytorch 1.5 version for Inference

* Adding new line at the eof

Co-authored-by: Tushar Dey &lt;tshdy@amazon.com&gt;
diff --git a/buildspec.yml b/buildspec.yml
@@ -2,12 +2,10 @@ version: 0.2
 
 env:
   variables:
-    FRAMEWORK_VERSION: '1.4.0'
+    FRAMEWORK_VERSION: '1.5.0'
     EIA_FRAMEWORK_VERSION: '1.3.1'
-    CPU_PY2_VERSION: '2'
     CPU_PY3_VERSION: '3'
     CPU_INSTANCE_TYPE: 'ml.c4.xlarge'
-    GPU_PY2_VERSION: '2'
     GPU_PY3_VERSION: '3'
     EIA_PY3_VERSION: '3'
     GPU_INSTANCE_TYPE: 'ml.p2.xlarge'
@@ -40,16 +38,6 @@ phases:
       - gpu_dockerfile="Dockerfile.gpu"
       - eia_dockerfile="Dockerfile.eia"
 
-      # build py2 images
-      - build_dir="docker/$FRAMEWORK_VERSION/py$CPU_PY2_VERSION"
-      - cp -r docker/build_artifacts/* $build_dir/
-      - CPU_PY2_TAG="$FRAMEWORK_VERSION-cpu-py2-$BUILD_ID"
-      - GPU_PY2_TAG="$FRAMEWORK_VERSION-gpu-py2-$BUILD_ID"
-      - build_cmd="docker build -f "$build_dir/$cpu_dockerfile" -t $PREPROD_IMAGE:$CPU_PY2_TAG $build_dir"
-      - execute-command-if-has-matching-changes "$build_cmd" "test/" "docker/*" "buildspec.yml"
-      - build_cmd="docker build -f "$build_dir/$gpu_dockerfile" -t $PREPROD_IMAGE:$GPU_PY2_TAG $build_dir"
-      - execute-command-if-has-matching-changes "$build_cmd" "test/" "docker/*" "buildspec.yml"
-
       # build py3 image
       - build_dir="docker/$FRAMEWORK_VERSION/py$GPU_PY3_VERSION"
       - cp -r docker/build_artifacts/* $build_dir/
@@ -67,12 +55,8 @@ phases:
 
       # push images to ecr
       - $(aws ecr get-login --registry-ids $ACCOUNT --no-include-email --region $AWS_DEFAULT_REGION)
-      - push_cmd="docker push $PREPROD_IMAGE:$CPU_PY2_TAG"
-      - execute-command-if-has-matching-changes "$push_cmd" "test/" "docker/*" "buildspec.yml"
       - push_cmd="docker push $PREPROD_IMAGE:$CPU_PY3_TAG"
       - execute-command-if-has-matching-changes "$push_cmd" "test/" "docker/*" "buildspec.yml"
-      - push_cmd="docker push $PREPROD_IMAGE:$GPU_PY2_TAG"
-      - execute-command-if-has-matching-changes "$push_cmd" "test/" "docker/*" "buildspec.yml"
       - push_cmd="docker push $PREPROD_IMAGE:$GPU_PY3_TAG"
       - execute-command-if-has-matching-changes "$push_cmd" "test/" "docker/*" "buildspec.yml"
       - push_cmd="docker push $PREPROD_IMAGE:$EIA_PY3_TAG"
@@ -86,31 +70,21 @@ phases:
 
       # run cpu integration tests
       - py3_cmd="pytest test/integration/local --region $AWS_DEFAULT_REGION --docker-base-name $PREPROD_IMAGE --framework-version $FRAMEWORK_VERSION --py-version $CPU_PY3_VERSION --processor cpu --tag $CPU_PY3_TAG"
-      - py2_cmd="pytest test/integration/local --region $AWS_DEFAULT_REGION --docker-base-name $PREPROD_IMAGE --framework-version $FRAMEWORK_VERSION --py-version $CPU_PY2_VERSION --processor cpu --tag $CPU_PY2_TAG"
       - execute-command-if-has-matching-changes "$py3_cmd" "test/" "docker/*" "buildspec.yml"
-      - execute-command-if-has-matching-changes "$py2_cmd" "test/" "docker/*" "buildspec.yml"
 
       # run gpu integration tests
       - printf "$SETUP_CMDS" > $SETUP_FILE
       - py3_pytest_cmd="pytest test/integration/local --region $AWS_DEFAULT_REGION --docker-base-name $PREPROD_IMAGE --framework-version $FRAMEWORK_VERSION --py-version $GPU_PY3_VERSION --processor gpu --tag $GPU_PY3_TAG"
       - py3_cmd="remote-test --github-repo $GITHUB_REPO --test-cmd \"$py3_pytest_cmd\" --setup-file $SETUP_FILE --pr-number \"$PR_NUM\""
       - execute-command-if-has-matching-changes "$py3_cmd" "test/" "docker/*" "buildspec.yml"
 
-      - py2_pytest_cmd="pytest test/integration/local --region $AWS_DEFAULT_REGION --docker-base-name $PREPROD_IMAGE --framework-version $FRAMEWORK_VERSION --py-version $GPU_PY2_VERSION --processor gpu --tag $GPU_PY2_TAG"
-      - py2_cmd="remote-test --github-repo $GITHUB_REPO --test-cmd \"$py2_pytest_cmd\" --setup-file $SETUP_FILE --pr-number \"$PR_NUM\" --skip-setup"
-      - execute-command-if-has-matching-changes "$py2_cmd" "test/" "docker/*" "buildspec.yml"
-
       # run cpu sagemaker tests
       - py3_cmd="pytest test/integration/sagemaker --region $AWS_DEFAULT_REGION --docker-base-name $ECR_REPO --aws-id $ACCOUNT --framework-version $FRAMEWORK_VERSION --py-version $CPU_PY3_VERSION --processor cpu --instance-type $CPU_INSTANCE_TYPE --tag $CPU_PY3_TAG"
-      - py2_cmd="pytest test/integration/sagemaker --region $AWS_DEFAULT_REGION --docker-base-name $ECR_REPO --aws-id $ACCOUNT --framework-version $FRAMEWORK_VERSION --py-version $CPU_PY2_VERSION --processor cpu --instance-type $CPU_INSTANCE_TYPE --tag $CPU_PY2_TAG"
       - execute-command-if-has-matching-changes "$py3_cmd" "test/" "docker/*" "buildspec.yml"
-      - execute-command-if-has-matching-changes "$py2_cmd" "test/" "docker/*" "buildspec.yml"
 
       # run gpu sagemaker tests
       - py3_cmd="pytest test/integration/sagemaker --region $AWS_DEFAULT_REGION --docker-base-name $ECR_REPO --aws-id $ACCOUNT --framework-version $FRAMEWORK_VERSION --py-version $GPU_PY3_VERSION --processor gpu --instance-type $GPU_INSTANCE_TYPE --tag $GPU_PY3_TAG"
-      - py2_cmd="pytest test/integration/sagemaker --region $AWS_DEFAULT_REGION --docker-base-name $ECR_REPO --aws-id $ACCOUNT --framework-version $FRAMEWORK_VERSION --py-version $GPU_PY2_VERSION --processor gpu --instance-type $GPU_INSTANCE_TYPE --tag $GPU_PY2_TAG"
       - execute-command-if-has-matching-changes "$py3_cmd" "test/" "docker/*" "buildspec.yml"
-      - execute-command-if-has-matching-changes "$py2_cmd" "test/" "docker/*" "buildspec.yml"
 
       # run eia sagemaker tests
       - py3_cmd="pytest test/integration/sagemaker --region $AWS_DEFAULT_REGION --docker-base-name $ECR_REPO --aws-id $ACCOUNT --framework-version $EIA_FRAMEWORK_VERSION --py-version $EIA_PY3_VERSION --processor cpu --instance-type $CPU_INSTANCE_TYPE --accelerator-type $EIA_ACCELERATOR_TYPE --tag $EIA_PY3_TAG"
@@ -122,8 +96,6 @@ phases:
       - cleanup-key-pairs
 
       # remove ecr image
-      - aws ecr batch-delete-image --repository-name $ECR_REPO --region $AWS_DEFAULT_REGION --image-ids imageTag=$CPU_PY2_TAG
       - aws ecr batch-delete-image --repository-name $ECR_REPO --region $AWS_DEFAULT_REGION --image-ids imageTag=$CPU_PY3_TAG
-      - aws ecr batch-delete-image --repository-name $ECR_REPO --region $AWS_DEFAULT_REGION --image-ids imageTag=$GPU_PY2_TAG
       - aws ecr batch-delete-image --repository-name $ECR_REPO --region $AWS_DEFAULT_REGION --image-ids imageTag=$GPU_PY3_TAG
       - aws ecr batch-delete-image --repository-name $ECR_REPO --region $AWS_DEFAULT_REGION --image-ids imageTag=$EIA_PY3_TAG
diff --git a/docker/1.5.0/py3/Dockerfile.cpu b/docker/1.5.0/py3/Dockerfile.cpu
@@ -0,0 +1,88 @@
+FROM ubuntu:16.04
+
+LABEL maintainer="Amazon AI"
+LABEL com.amazonaws.sagemaker.capabilities.accept-bind-to-port=true
+LABEL com.amazonaws.sagemaker.capabilities.multi-models=true
+
+ARG PYTHON_VERSION=3.6.6
+ARG PYTORCH_VERSION=1.5.0
+ARG TORCHVISION_VERSION=0.6.0
+ARG MMS_VERSION=1.0.8
+
+# See http://bugs.python.org/issue19846
+ENV LANG C.UTF-8
+ENV LD_LIBRARY_PATH /opt/conda/lib/:$LD_LIBRARY_PATH
+ENV PATH /opt/conda/bin:$PATH
+ENV SAGEMAKER_SERVING_MODULE sagemaker_pytorch_serving_container.serving:main
+ENV TEMP=/home/model-server/tmp
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    ca-certificates \
+    cmake \
+    curl \
+    git \
+    jq \
+    libgl1-mesa-glx \
+    libglib2.0-0 \
+    libsm6 \
+    libxext6 \
+    libxrender-dev \
+    openjdk-8-jdk-headless \
+    vim \
+    wget \
+    zlib1g-dev
+
+RUN curl -L -o ~/miniconda.sh https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh \
+ && chmod +x ~/miniconda.sh \
+ && ~/miniconda.sh -b -p /opt/conda \
+ && rm ~/miniconda.sh \
+ && /opt/conda/bin/conda update conda \
+ && /opt/conda/bin/conda install -y \
+    python=$PYTHON_VERSION \
+    cython==0.29.12 \
+    ipython==7.7.0 \
+    mkl-include==2019.4 \
+    mkl==2019.4 \
+    numpy==1.16.4 \
+    scipy==1.3.0 \
+    typing==3.6.4 \
+ && /opt/conda/bin/conda clean -ya
+
+RUN conda install -c \
+    conda-forge \
+    opencv==4.0.1 \
+ && conda install -y \
+    scikit-learn==0.21.2 \
+    pandas==0.25.0 \
+    h5py==2.9.0 \
+    requests==2.22.0 \
+ && conda clean -ya \
+ && pip install --upgrade pip --trusted-host pypi.org --trusted-host files.pythonhosted.org \
+ && ln -s /opt/conda/bin/pip /usr/local/bin/pip3 \
+ && pip install mxnet-model-server==$MMS_VERSION
+
+ # Uninstall and re-install torch and torchvision from the PyTorch website
+RUN pip uninstall -y torch \
+ && pip uninstall -y torchvision \
+ && pip install torch==$PYTORCH_VERSION+cpu torchvision==$TORCHVISION_VERSION+cpu -f https://download.pytorch.org/whl/torch_stable.html
+
+RUN useradd -m model-server \
+ && mkdir -p /home/model-server/tmp \
+ && chown -R model-server /home/model-server
+
+COPY mms-entrypoint.py /usr/local/bin/dockerd-entrypoint.py
+COPY config.properties /home/model-server
+
+RUN chmod +x /usr/local/bin/dockerd-entrypoint.py
+
+RUN pip install --no-cache-dir "sagemaker-pytorch-inference<2"
+
+RUN curl https://aws-dlc-licenses.s3.amazonaws.com/pytorch-1.5.0/license.txt -o /license.txt
+
+RUN conda install -y -c conda-forge pyyaml==5.3.1
+RUN pip install  sagemaker-containers==2.8.6 pillow==7.1.0 awscli
+
+EXPOSE 8080 8081
+ENTRYPOINT ["python", "/usr/local/bin/dockerd-entrypoint.py"]
+CMD ["mxnet-model-server", "--start", "--mms-config", "/home/model-server/config.properties"]
diff --git a/docker/1.5.0/py3/Dockerfile.gpu b/docker/1.5.0/py3/Dockerfile.gpu
@@ -0,0 +1,103 @@
+FROM nvidia/cuda:10.1-cudnn7-devel-ubuntu16.04
+# NCCL_VERSION=2.4.7, CUDNN_VERSION=7.6.2.24
+LABEL maintainer="Amazon AI"
+LABEL com.amazonaws.sagemaker.capabilities.accept-bind-to-port=true
+
+# Add arguments to achieve the version, python and url
+ARG PYTHON_VERSION=3.6.6
+ARG PYTORCH_VERSION=1.5.0
+ARG TORCHVISION_VERSION=0.6.0
+ARG MMS_VERSION=1.0.8
+
+# See http://bugs.python.org/issue19846
+ENV LANG C.UTF-8
+ENV LD_LIBRARY_PATH /opt/conda/lib/:$LD_LIBRARY_PATH
+ENV PATH /opt/conda/bin:$PATH
+ENV SAGEMAKER_SERVING_MODULE sagemaker_pytorch_serving_container.serving:main
+ENV TEMP=/home/model-server/tmp
+
+RUN apt-get update \
+ && apt-get install -y --allow-downgrades --allow-change-held-packages --no-install-recommends \
+    build-essential \
+    ca-certificates \
+    cmake \
+    curl \
+    git \
+    jq \
+    libgl1-mesa-glx \
+    libglib2.0-0 \
+    libgomp1 \
+    libibverbs-dev \
+    libsm6 \
+    libxext6 \
+    libxrender-dev \
+    openjdk-8-jdk-headless \
+    vim \
+    wget \
+    zlib1g-dev
+
+# Install OpenSSH. Allow OpenSSH to talk to containers without asking for confirmation
+RUN apt-get install -y --no-install-recommends \
+    openssh-client \
+    openssh-server \
+ && mkdir -p /var/run/sshd \
+ && cat /etc/ssh/ssh_config | grep -v StrictHostKeyChecking > /etc/ssh/ssh_config.new \
+ && echo "    StrictHostKeyChecking no" >> /etc/ssh/ssh_config.new \
+ && mv /etc/ssh/ssh_config.new /etc/ssh/ssh_configs
+
+RUN curl -L -o ~/miniconda.sh https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh \
+ && chmod +x ~/miniconda.sh \
+ && ~/miniconda.sh -b -p /opt/conda \
+ && rm ~/miniconda.sh \
+ && /opt/conda/bin/conda update conda \
+ && /opt/conda/bin/conda install -y \
+    python=$PYTHON_VERSION \
+    cython==0.29.12 \
+    ipython==7.7.0 \
+    mkl-include==2019.4 \
+    mkl==2019.4 \
+    numpy==1.16.4 \
+    scipy==1.3.0 \
+    typing==3.6.4 \
+ && /opt/conda/bin/conda clean -ya
+
+RUN conda install -c \
+    pytorch magma-cuda101 \
+ && conda install -c \
+    conda-forge \
+    opencv==4.0.1 \
+ && conda install -y \
+    scikit-learn==0.21.2 \
+    pandas==0.25.0 \
+    h5py==2.9.0 \
+    requests==2.22.0 \
+ && conda clean -ya \
+ && /opt/conda/bin/conda config --set ssl_verify False \
+ && pip install --upgrade pip --trusted-host pypi.org --trusted-host files.pythonhosted.org \
+ && ln -s /opt/conda/bin/pip /usr/local/bin/pip3 \
+ && pip install mxnet-model-server==$MMS_VERSION
+
+# Uninstall and re-install torch and torchvision from the PyTorch website
+RUN pip uninstall -y torch \
+ && pip uninstall -y torchvision \
+ && pip install torch==$PYTORCH_VERSION+cu101 torchvision==$TORCHVISION_VERSION+cu101 -f https://download.pytorch.org/whl/torch_stable.html
+
+RUN useradd -m model-server \
+ && mkdir -p /home/model-server/tmp \
+ && chown -R model-server /home/model-server
+
+COPY mms-entrypoint.py /usr/local/bin/dockerd-entrypoint.py
+COPY config.properties /home/model-server
+
+RUN chmod +x /usr/local/bin/dockerd-entrypoint.py
+
+RUN pip install --no-cache-dir "sagemaker-pytorch-inference<2"
+
+RUN curl https://aws-dlc-licenses.s3.amazonaws.com/pytorch-1.5.0/license.txt -o /license.txt
+
+RUN conda install -y -c conda-forge pyyaml==5.3.1
+RUN pip install  sagemaker-containers==2.8.6 pillow==7.1.0 awscli
+
+EXPOSE 8080 8081
+ENTRYPOINT ["python", "/usr/local/bin/dockerd-entrypoint.py"]
+CMD ["mxnet-model-server", "--start", "--mms-config", "/home/model-server/config.properties"]
