tensorflow-deployment.yml

ROSTemplateFormatVersion: '2015-09-01'
Description:
  zh-cn: 创建ECS实例群，配置VPC与安全组，自动安装TensorFlow，支持GPU选项和网络流量控制，提供实例公网IP和访问安全。
  en: Create an ECS instance cluster, configure VPC and security groups, automate
    the installation of TensorFlow, with support for GPU options and network traffic
    control, providing public IP addresses for instances and ensuring secure access.
Conditions:
  gpu_disable:
    Fn::Equals:
    - None
    - Ref: GPUNodeZoneId
Parameters:
  EcsAmount:
    Type: Number
    Label:
      en: Nodes Amount
      zh-cn: 实例的数量
    Description:
      en: 'The number of ECS instances will be created,Allowed value: 1~100.'
      zh-cn: 创建实例的数量，范围1-100。
    Default: 1
    MinValue: 1
    MaxValue: 100
  GPUNodeZoneId:
    Type: String
    Label:
      en: Zone of GPU Node
      zh-cn: GPU节点区域
    Description:
      en: Specify zone Id that provides the GPU instance while InstanceType is specified
        the GPU type.
      zh-cn: 指定GPU实例的可用区ID，并确定GPU规格库存情况。
    Default: None
  EcsInstanceType:
    Type: String
    Label:
      en: Instance Type
      zh-cn: 实例规格
    Description:
      en: <font color='blue'><b>1.Before selecting the model please confirm that the
        current available zone under the model is in stock, some models need to be
        reported in advance</b></font><br><font color='blue'><b>2.List of optional
        models</font><br></b></font>[ecs.c5.large <font color='green'>2vCPU 4GiB Intranet
        bandwidth1Gbps In-grid sending and receiving packages30MillionPPS</font>]<br></b>[ecs.c5.xlarge
        <font color='green'>4vCPU 8GiB Intranet bandwidth1.5Gbps In-grid sending and
        receiving packages50MillionPPS</font>]<br></b>[ecs.c5.2xlarge <font color='green'>8vCPU
        16GiB Intranet bandwidth2.5Gbps In-grid sending and receiving packages80MillionPPS</font>]
      zh-cn: <font color='blue'><b>1.选择机型前请先确认当前可用区下该机型是否有货，部分机型需要提前报备</b></font><br><font
        color='blue'><b>2.可选机型列表</font><br></b></font>[ecs.c5.large <font color='green'>2vCPU
        4GiB 内网带宽1Gbps 内网收发包30万PPS</font>]<br></b>[ecs.c5.xlarge <font color='green'>4vCPU
        8GiB 内网带宽1.5Gbps 内网收发包50万PPS</font>]<br></b>[ecs.c5.2xlarge <font color='green'>8vCPU
        16GiB 内网带宽2.5Gbps 内网收发包80万PPS</font>]
    AssociationProperty: ALIYUN::ECS::Instance::InstanceType
    AssociationPropertyMetadata:
      ZoneId: ZoneId
  EcsImageId:
    Type: String
    Label:
      en: Image ID
      zh-cn: 镜像ID
    Description:
      en: 'Instance Image ID. see detail: <b><a href=''https://www.alibabacloud.com/help/doc-detail/112977.html''
        target=''_blank''><font color=''blue''>Find the mirror</font></a></b>'
      zh-cn: 实例镜像，详见：<b><a href='https://help.aliyun.com/document_detail/112977.html'
        target='_blank'><font color='blue'>查找镜像</font></a></b>
    AssociationProperty: ALIYUN::ECS::Image::ImageId
    AssociationPropertyMetadata:
      InstanceType: ${EcsInstanceType}
      SupportedImageOwnerAlias:
        - system
        - self
        - others
    Default: centos_7_04_64_20G_alibase_201701015.vhd
  EcsPassword:
    Type: String
    Label:
      en: Instance Password
      zh-cn: 实例密码
    Description:
      en: The 8-30 long login password of instance, consists of the uppercase, lowercase
        letter and number. <br> special characters include()`~!@#$%^&*_-+=|{}[]:;'<>,.?/
      zh-cn: 长度8-30，必须包含大写字母、小写字母、数字、特殊符号三个；<br>特殊字符包括：()`~!@#$%^&*_-+=|{}[]:;'<>,.?/
    ConstraintDescription:
      en: 'Length 8-30, must contain upper case letters, lower case letters, Numbers,
        special symbols three; special characters include: ()`~!@#$%^&*_-+=|{}[]:;''<>,.?/'
      zh-cn: 长度8-30，必须包含大写字母、小写字母、数字、特殊符号三种；特殊字符包括：()`~!@#$%^&*_-+=|{}[]:;' <>,.?/
    AllowedPattern: '[0-9A-Za-z\_\-&:;''<>,=%`~!@#\(\)\$\^\*\+\|\{\}\[\]\.\?\/]+$'
    MinLength: '8'
    MaxLength: '30'
    NoEcho: true
  ZoneId:
    Type: String
    Label:
      en: Zone ID
      zh-cn: 可用区ID
    AssociationProperty: ALIYUN::ECS::Instance::ZoneId
Resources:
  EcsVpc:
    Type: ALIYUN::ECS::VPC
    Properties:
      CidrBlock: 192.168.0.0/16
      VpcName: TensorFlow_Vpc
  EcsSecurityGroup:
    Type: ALIYUN::ECS::SecurityGroup
    Properties:
      VpcId:
        Ref: EcsVpc
      SecurityGroupEgress:
      - DestCidrIp: 0.0.0.0/0
        IpProtocol: all
        NicType: intranet
        PortRange: -1/-1
        Priority: 1
      SecurityGroupIngress:
      - IpProtocol: tcp
        NicType: intranet
        PortRange: 6006/6006
        Priority: 1
        SourceCidrIp: 0.0.0.0/0
      SecurityGroupName: TensorFlow_SG
  EcsVSwitch:
    Type: ALIYUN::ECS::VSwitch
    Properties:
      ZoneId:
        Fn::If:
        - gpu_disable
        - Ref: ZoneId
        - Ref: GPUNodeZoneId
      VpcId:
        Ref: EcsVpc
      CidrBlock: 192.168.0.0/16
      VSwitchName: TensorFlow_VSwitch
  EcsInstanceGroup:
    Type: ALIYUN::ECS::InstanceGroup
    Properties:
      VpcId:
        Ref: EcsVpc
      VSwitchId:
        Ref: EcsVSwitch
      SecurityGroupId:
        Ref: EcsSecurityGroup
      ImageId:
        Ref: EcsImageId
      AllocatePublicIP: true
      InstanceType:
        Ref: EcsInstanceType
      InternetMaxBandwidthOut: 50
      MaxAmount:
        Ref: EcsAmount
      Password:
        Ref: EcsPassword
      UserData:
        Fn::Replace:
        - ros-notify:
            Fn::GetAtt:
            - RosWaitConditionHandle
            - CurlCli
        - Fn::Join:
          - ''
          - - '#!/bin/sh

              '
            - gpu_enable=
            - Fn::If:
              - gpu_disable
              - 'no'
              - 'yes'
            - '

              '
            - 'apt-get install -y python-pip python-dev

              '
            - 'pip install tensorflow

              '
            - 'if [ $gpu_enable = ''yes'' ]; then

              '
            - '    pip install tensorflow-gpu

              '
            - '    wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_8.0.61-1_amd64.deb

              '
            - '    dpkg -i cuda-repo-ubuntu1604_8.0.61-1_amd64.deb

              '
            - '    apt-get install -y libgd3

              '
            - '    apt-get update --fix-missing

              '
            - '    apt-get install -y cuda

              '
            - '    echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64
              /" > /etc/apt/sources.list.d/nvidia-ml.list

              '
            - '    apt-get update

              '
            - '    apt-get install -y --no-install-recommends libcudnn7=7.0.1.13-1+cuda8.0

              '
            - '    apt-get install -y --no-install-recommends libcudnn7-dev=7.0.1.13-1+cuda8.0

              '
            - '    cd /usr/local/cuda/lib64

              '
            - '    ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so.7.0.1 libcudnn.so

              '
            - '    ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so.7.0.1 libcudnn.so.6

              '
            - '    cp /usr/lib/x86_64-linux-gnu/libcudnn_static_v7.a .

              '
            - '    cd -

              '
            - '    echo export CUDA_HOME=/usr/local/cuda > /root/.bash_profile

              '
            - '    echo ''export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:$CUDA_HOME/lib64:/usr/local/cuda/extras/CUPTI/lib64"''
              >> /root/.bash_profile

              '
            - '    echo ''export PATH="$CUDA_HOME/bin:$PATH"'' >>  /root/.bash_profile

              '
            - '    source /root/.bash_profile

              '
            - '    export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64"

              '
            - "    export CUDA_HOME=/usr/local/cuda \n"
            - 'fi

              '
            - 'tensorboard --logdir=/tmp/tf_logs &

              '
            - "ros-notify \n"
  RosWaitConditionHandle:
    Type: ALIYUN::ROS::WaitConditionHandle
    Properties: {}
  RosWaitCondition:
    Type: ALIYUN::ROS::WaitCondition
    Properties:
      Count:
        Ref: EcsAmount
      Handle:
        Ref: RosWaitConditionHandle
      Timeout: 1800
Outputs:
  tensorflow_nodes_ips:
    Description:
      en: Public IP address list of created ecs instance.
      zh-cn: 已创建的ecs实例的公共IP地址列表。
    Value:
      Fn::GetAtt:
      - EcsInstanceGroup
      - PublicIps
Metadata:
  ALIYUN::ROS::Interface:
    ParameterGroups:
    - Parameters:
      - ZoneId
      - EcsAmount
      - GPUNodeZoneId
      - EcsInstanceType
      - EcsImageId
      - EcsPassword
      Label:
        default:
          en: ECS
          zh-cn: 云服务器
    TemplateTags:
    - acs:example:人工智能:部署TensorFlow开发环境
  ALIYUN::ROS::Composer:
    '40402546':
      Res:
        - EcsVpc
      Parent: 456de6cb
      Rect:
        - 480
        - 340
        - 80
        - 200
        - 3
        - 0
    c0994419:
      Rect:
        - 620
        - 540
        - 40
        - 100
        - 1
        - 0
      ResT: Composer::ROSParameter::AlibabaCloud
    456de6cb:
      Parent: c0994419
      Rect:
        - 580
        - 470
        - 60
        - 150
        - 2
        - 0
      ResT: Composer::ROSParameter::Region
    eb43794c:
      Res:
        - RosWaitConditionHandle
      Parent: 456de6cb
      Rect:
        - 40
        - 40
        - 206
        - 540
        - 3
        - 0
    c19bfb74:
      Res:
        - RosWaitCondition
      Parent: 456de6cb
      Rect:
        - 40
        - 40
        - 385
        - 540
        - 3
        - 0
    79c6f97f:
      Res:
        - ZoneId
      Parent: '40402546'
      Rect:
        - 440
        - 270
        - 100
        - 250
        - 4
        - 0
      ResT: Composer::ROSParameter::Zone
    897c79ac:
      Res:
        - EcsVSwitch
      Parent: 79c6f97f
      Rect:
        - 400
        - 200
        - 120
        - 300
        - 5
        - 0
    3df1f4b9:
      Res:
        - EcsSecurityGroup
      Parent: '40402546'
      Rect:
        - 165
        - 130
        - 233
        - 341
        - 9
        - 0
    d3d3d8c9:
      Res:
        - EcsInstanceGroup
      Parent: 897c79ac
      Rect:
        - 40
        - 40
        - 300
        - 380
        - 10
        - 0
      Layer:
        - 3df1f4b9
    ce1b3e69:
      Parent: 456de6cb
      Edge:
        - c19bfb74
        - eb43794c
      Line: 0:0:0:gray:0