AWSTemplateFormatVersion: 2010-09-09
Description: ECS on spot instances example
Mappings:
  ecsoptimizedami:
    ap-northeast-1:
      ami: ami-f63f6f91
    ap-southeast-1:
      ami: ami-b4ae1dd7
    ap-southeast-2:
      ami: ami-fbe9eb98
    ca-central-1:
      ami: ami-ee58e58a
    eu-central-1:
      ami: ami-085e8a67
    eu-west-1:
      ami: ami-95f8d2f3
    eu-west-2:
      ami: ami-bf9481db
    us-east-1:
      ami: ami-275ffe31
    us-east-2:
      ami: ami-62745007
    us-west-1:
      ami: ami-689bc208
    us-west-2:
      ami: ami-62d35c02
Metadata:
  Authors:
    Description: anshrma@amazon.com
  License:
    Description: 'Copyright 2017 Amazon.com, Inc. and its affiliates. All Rights Reserved.
    SPDX-License-Identifier: MIT-0'
Parameters:

  SpotBidPrice:
    Default: 0.05
    Description: Spot Instance bid price
    Type: String
  InstanceType:
    Default: c3.large
    Description: Instance Type
    Type: String

  InstanceProfile:
    Description: InstanceProfile
    Type: String
  InstanceSG:
    Description: Security Groups on Instance
    Type: AWS::EC2::SecurityGroup::Id
  PurchaseOption:
    Description: Security Groups on Instance
    Type: String
    AllowedValues:
      - OnDemand
      - Spot
  CloudWatchLogsGroup:
    Description: Log Group for log aggregation
    Type: String
  Subnet1:
    Description: Subnet where instances will be placed
    Type: AWS::EC2::Subnet::Id
  Subnet2:
    Description: Subnets where instances will be placed
    Type: AWS::EC2::Subnet::Id
  ServiceRole:
    Description: Role used by ECS Service
    Type: String
  TaskRole:
    Description: Role used by ECS Tasks
    Type: String
  TargetGroupARN:
    Description: ARN of the Target Group
    Type: String
  snsTopic:
    Description: SNS Topic where termination notices will be sent
    Type: String
  MinTaskNumber:
    Description: Min. Number of Tasks the Service should run
    Type: String
  MaxTaskNumber:
    Description: Max. Number of Tasks the Service should run
    Type: String
  TaskScaleUpAdjustment:
    Description: The number by which the tasks should increment
    Type: String
  TaskScaleDownAdjustment:
    Description: The number by which the tasks should decrease
    Type: String


Conditions:
  IsSpot: !Equals [ !Ref PurchaseOption, Spot ]
Resources:

  ecsCluster:
    Type: AWS::ECS::Cluster
  LaunchConfiguration:
    Type: AWS::AutoScaling::LaunchConfiguration
    Metadata:
      AWS::CloudFormation::Init:
        config:
          files:
            "/etc/cfn/cfn-hup.conf":
              mode: 000400
              owner: root
              group: root
              content: !Sub |
                [main]
                stack=${AWS::StackId}
                region=${AWS::Region}
            "/etc/cfn/hooks.d/cfn-auto-reloader.conf":
              content: !Sub |
                [cfn-auto-reloader-hook]
                triggers=post.update
                path=Resources.ContainerInstances.Metadata.AWS::CloudFormation::Init
                action=/opt/aws/bin/cfn-init -v --region ${AWS::Region} --stack ${AWS::StackName} --resource LaunchConfiguration
          services:
            sysvinit:
              cfn-hup:
                enabled: true
                ensureRunning: true
                files:
                  - /etc/cfn/cfn-hup.conf
                  - /etc/cfn/hooks.d/cfn-auto-reloader.conf
    Properties:
      ImageId: !FindInMap [ecsoptimizedami,!Ref "AWS::Region", ami]
      InstanceType: !Ref InstanceType
      IamInstanceProfile: !Ref InstanceProfile
      SpotPrice: !If [IsSpot,!Ref SpotBidPrice,!Ref "AWS::NoValue"]
      SecurityGroups:
        - !Ref InstanceSG
      UserData:
            Fn::Base64: !Sub |
              #!/bin/bash -xe
              export PATH=/usr/local/bin:$PATH
              yum -y --security update
              yum -y install jq
              yum install -y aws-cfn-bootstrap
              easy_install pip
              pip install awscli
              aws configure set default.region ${AWS::Region}
              echo ECS_CLUSTER=${ecsCluster} >> /etc/ecs/ecs.config
              echo ECS_ENABLE_TASK_IAM_ROLE=true >> /etc/ecs/ecs.config
              echo ECS_INSTANCE_ATTRIBUTES={\"instance-purchase-option\":\"${PurchaseOption}\"} >> /etc/ecs/ecs.config

              cat <<EOF > /tmp/awslogs.conf

              [general]
              state_file = /var/awslogs/state/agent-state

              [/var/log/dmesg]
              file = /var/log/dmesg
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/dmesg
              initial_position = start_of_file

              [/var/log/messages]
              file = /var/log/messages
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/messages
              datetime_format = %b %d %H:%M:%S
              initial_position = start_of_file

              [/var/log/docker]
              file = /var/log/docker
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/docker
              datetime_format = %Y-%m-%dT%H:%M:%S.%f
              initial_position = start_of_file


              [/var/log/ecs/ecs-init.log]
              file = /var/log/ecs/ecs-init.log.*
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/ecs/ecs-init.log
              datetime_format = %Y-%m-%dT%H:%M:%SZ
              initial_position = start_of_file


              [/var/log/ecs/ecs-agent.log]
              file = /var/log/ecs/ecs-agent.log.*
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/ecs/ecs-agent.log
              datetime_format = %Y-%m-%dT%H:%M:%SZ
              initial_position = start_of_file


              [/var/log/ecs/audit.log]
              file = /var/log/ecs/audit.log.*
              log_group_name = ${CloudWatchLogsGroup}
              log_stream_name = %ECS_CLUSTER/%CONTAINER_INSTANCE/var/log/ecs/audit.log
              datetime_format = %Y-%m-%dT%H:%M:%SZ
              initial_position = start_of_file
              EOF


              cd /tmp && curl -sO https://s3.amazonaws.com/aws-cloudwatch/downloads/latest/awslogs-agent-setup.py
              python /tmp/awslogs-agent-setup.py -n -r ${AWS::Region} -c /tmp/awslogs.conf


              cat <<EOF > /etc/init/cloudwatch-logs-start.conf
              description "Configure and start CloudWatch Logs agent on Amazon ECS containerinstance"
              author "Amazon Web Services"
              start on started ecs
              script
              exec 2>>/var/log/cloudwatch-logs-start.log
              set -x
              until curl -s http://localhost:51678/v1/metadata; do sleep 1; done
              ECS_CLUSTER=\$(curl -s http://localhost:51678/v1/metadata | jq .Cluster | tr-d \")
              CONTAINER_INSTANCE=\$(curl -s http://localhost:51678/v1/metadata | jq .ContainerInstanceArn| tr -d \")
              sed -i "s|%ECS_CLUSTER|\$ECS_CLUSTER|g" /var/awslogs/etc/awslogs.conf
              sed -i "s|%CONTAINER_INSTANCE|\$CONTAINER_INSTANCE|g" /var/awslogs/etc/awslogs.conf
              chkconfig awslogs on
              service awslogs start
              end script
              EOF


              cat <<EOF > /etc/init/spot-instance-termination-notice-handler.conf
              description "Start spot instance termination handler monitoring script"
              author "Amazon Web Services"
              start on started ecs
              script
              echo \$\$ > /var/run/spot-instance-termination-notice-handler.pid
              exec /usr/local/bin/spot-instance-termination-notice-handler.sh
              end script
              pre-start script
              logger "[spot-instance-termination-notice-handler.sh]: spot instance terminationnotice handler started"
              end script
              EOF


              cat <<EOF > /usr/local/bin/spot-instance-termination-notice-handler.sh
              #!/bin/bash
              while sleep 5; do
              if [ -z \$(curl -Isf http://169.254.169.254/latest/meta-data/spot/termination-time)];then
              /bin/false
              else
              logger "[spot-instance-termination-notice-handler.sh]: spot instance terminationnotice detected"
              STATUS=DRAINING
              ECS_CLUSTER=\$(curl -s http://localhost:51678/v1/metadata | jq .Cluster | tr-d \")
              CONTAINER_INSTANCE=\$(curl -s http://localhost:51678/v1/metadata | jq .ContainerInstanceArn| tr -d \")
              logger "[spot-instance-termination-notice-handler.sh]: putting instance in state\$STATUS"
              logger "[spot-instance-termination-notice-handler.sh]: running: /usr/local/bin/awsecs update-container-instances-state --cluster \$ECS_CLUSTER --container-instances\$CONTAINER_INSTANCE --status \$STATUS"
              /usr/local/bin/aws ecs update-container-instances-state --cluster \$ECS_CLUSTER--container-instances \$CONTAINER_INSTANCE --status \$STATUS
              logger "[spot-instance-termination-notice-handler.sh]: running: \"/usr/local/bin/awssns publish --topic-arn ${snsTopic} --message \"Spot instance termination noticedetected. Details: cluster: \$ECS_CLUSTER, container_instance: \$CONTAINER_INSTANCE.Putting instance in state \$STATUS.\""
              /usr/local/bin/aws sns publish --topic-arn ${snsTopic} --message "Spot instancetermination notice detected. Details: cluster: \$ECS_CLUSTER, container_instance:\$CONTAINER_INSTANCE. Putting instance in state \$STATUS."
              logger "[spot-instance-termination-notice-handler.sh]: putting myself to sleep..."
              sleep 120
              fi
              done
              EOF


              chmod +x /usr/local/bin/spot-instance-termination-notice-handler.sh
              /opt/aws/bin/cfn-init -v --region ${AWS::Region} --stack ${AWS::StackName} --resource LaunchConfiguration
              /opt/aws/bin/cfn-signal -e $? --region ${AWS::Region} --stack ${AWS::StackName} --resource AutoScalingGroup
  AutoScalingGroup:
    Type: AWS::AutoScaling::AutoScalingGroup
    Properties:
      VPCZoneIdentifier:
        - !Ref Subnet1
        - !Ref Subnet2
      LaunchConfigurationName: !Ref LaunchConfiguration
      MinSize: 1
      MaxSize: 10
      DesiredCapacity: 1
      Tags:
        - Key: Name
          Value: !Sub ${AWS::StackName} - ECS Host
          PropagateAtLaunch: true
        - Key: Pricing
          Value: !Ref PurchaseOption
          PropagateAtLaunch: true


    CreationPolicy:
      ResourceSignal:
        Timeout: PT15M
    UpdatePolicy:
      AutoScalingRollingUpdate:
        MinInstancesInService: 1
        MaxBatchSize: 1
        PauseTime: PT15M
        WaitOnResourceSignals: true

  ECSService:
    Type: AWS::ECS::Service
    Properties:
      Cluster: !Ref ecsCluster
      Role: !Ref ServiceRole
      DesiredCount: !Ref MinTaskNumber
      TaskDefinition: !Ref TaskDefinition
      PlacementStrategies:
        - Type: spread
          Field: attribute:ecs.availability-zone
        - Type: spread
          Field: attribute:instance-purchase-option
      PlacementConstraints:
        - Type: memberOf
          Expression: !Sub attribute:instance-purchase-option == ${PurchaseOption}
      LoadBalancers:
        - ContainerName: nginx
          ContainerPort: 80
          TargetGroupArn: !Ref TargetGroupARN
  TaskDefinition:
    Type: AWS::ECS::TaskDefinition
    Properties:
      #Family: !Sub ${AWS::StackName}-simple-app
      ContainerDefinitions:
        - Name: nginx
          Image: nginx
          Essential: true
          Memory: 256
          Cpu: 256
          PortMappings:
            - ContainerPort: 80
              HostPort: 0
          LogConfiguration:
            LogDriver: awslogs
            Options:
              awslogs-group: !Ref CloudWatchLogsGroup
              awslogs-region: !Ref AWS::Region
              awslogs-stream-prefix: nginx
      Volumes: []
      TaskRoleArn: !Ref TaskRole

# Task Level AutoScaling Rules Starts here
  AutoScalingService:
    Type: AWS::ApplicationAutoScaling::ScalableTarget
    Properties:
      MaxCapacity: !Ref MaxTaskNumber
      MinCapacity: !Ref MinTaskNumber
      ResourceId: !Join [/, [service, !Ref ecsCluster, !GetAtt [ECSService, Name]]]
      RoleARN: !Ref ServiceRole
      ScalableDimension: ecs:service:DesiredCount
      ServiceNamespace: ecs
  ScalingUpPolicyService:
    Type: AWS::ApplicationAutoScaling::ScalingPolicy
    Properties:
      PolicyName: !Sub ${AWS::StackName}-${ECSService}-ScalingUpPolicy
      PolicyType: StepScaling
      ScalingTargetId: !Ref AutoScalingService
      StepScalingPolicyConfiguration:
        AdjustmentType: ChangeInCapacity
        Cooldown: 60
        StepAdjustments:
        - ScalingAdjustment: !Ref TaskScaleUpAdjustment
          MetricIntervalLowerBound: 0
  ScalingDownPolicyService:
    Type: AWS::ApplicationAutoScaling::ScalingPolicy
    Properties:
      PolicyName: !Sub ${AWS::StackName}-${ECSService}-ScalingDownPolicy
      PolicyType: StepScaling
      ScalingTargetId: !Ref AutoScalingService
      StepScalingPolicyConfiguration:
        AdjustmentType: ChangeInCapacity
        Cooldown: 60
        StepAdjustments:
        - ScalingAdjustment: !Ref TaskScaleDownAdjustment
          MetricIntervalUpperBound: -20
        - ScalingAdjustment: -1
          MetricIntervalLowerBound: -20
          MetricIntervalUpperBound: -10
  HighCPUServiceAlarm:
    Type: AWS::CloudWatch::Alarm
    Properties:
      AlarmDescription: CPUUtilization exceeding threshold . Triggers scale up
      ActionsEnabled: true
      Namespace: AWS/ECS
      MetricName: CPUUtilization
      Unit: Percent
      Dimensions:
      - Name: ClusterName
        Value: !Ref ecsCluster
      - Name: ServiceName
        Value: !GetAtt [ECSService, Name]
      Statistic: Average
      Period: '60'
      EvaluationPeriods: '2'
      Threshold: '70'
      AlarmActions: [!Ref ScalingUpPolicyService]
      ComparisonOperator: GreaterThanThreshold
  LowCPUServiceAlarm:
    Type: AWS::CloudWatch::Alarm
    Properties:
      AlarmDescription: CPUUtilization lowers threshold . Triggers scale down
      ActionsEnabled: true
      Namespace: AWS/ECS
      MetricName: CPUUtilization
      Unit: Percent
      Dimensions:
      - Name: ClusterName
        Value: !Ref ecsCluster
      - Name: ServiceName
        Value: !GetAtt [ECSService, Name]
      Statistic: Average
      Period: '60'
      EvaluationPeriods: '2'
      Threshold: '20'
      AlarmActions: [!Ref ScalingDownPolicyService]
      ComparisonOperator: LessThanThreshold


# Task Level AutoScaling Rules Ends here

# Cluster level AutoScaling Rules Start here

  ClusterScaleUpPolicy:
    Type: AWS::AutoScaling::ScalingPolicy
    Properties:
      PolicyType: SimpleScaling
      AdjustmentType: ChangeInCapacity
      AutoScalingGroupName: !Ref AutoScalingGroup
      Cooldown: 10
      ScalingAdjustment: 1
  ClusterScaleDownPolicy:
    Type: AWS::AutoScaling::ScalingPolicy
    Properties:
      PolicyType: SimpleScaling
      AdjustmentType: ChangeInCapacity
      AutoScalingGroupName: !Ref AutoScalingGroup
      Cooldown: 10
      ScalingAdjustment: -1
  ClusterHighMemoryUtilizationAlarm:
    Type: AWS::CloudWatch::Alarm
    Properties:
      AlarmDescription: Average Memory Utilization for the boxes in the ASG is above
        70% for 1 minutes. Triggers scale up
      ActionsEnabled: true
      Namespace: AWS/ECS
      MetricName: MemoryUtilization
      Unit: Percent
      Dimensions:
      - Name: ClusterName
        Value: !Ref ecsCluster
      Statistic: Maximum
      Period: '60'
      EvaluationPeriods: '1'
      Threshold: '70'
      AlarmActions: [!Ref ClusterScaleUpPolicy]
      ComparisonOperator: GreaterThanThreshold
  ClusterHighMemoryReservationAlarm:
    Type: AWS::CloudWatch::Alarm
    Properties:
      AlarmDescription: Average Memory Reservation for the boxes in the ASG is above
        70% for 3 minutes. Triggers scale up
      ActionsEnabled: true
      Namespace: AWS/ECS
      MetricName: MemoryReservation
      Unit: Percent
      Dimensions:
      - Name: ClusterName
        Value: !Ref ecsCluster
      Statistic: Maximum
      Period: '60'
      EvaluationPeriods: '1'
      Threshold: '70'
      AlarmActions: [!Ref ClusterScaleUpPolicy]
      ComparisonOperator: GreaterThanThreshold
  ClusterLowMemoryReservationAlarm:
    Type: AWS::CloudWatch::Alarm
    Properties:
      AlarmDescription: Average Memory Reservation for the boxes in the ASG is less.
        Triggers scale down
      ActionsEnabled: true
      Namespace: AWS/ECS
      MetricName: MemoryReservation
      Unit: Percent
      Dimensions:
      - Name: ClusterName
        Value: !Ref ecsCluster
      Statistic: Average
      Period: '60'
      EvaluationPeriods: '10'
      Threshold: '20'
      AlarmActions: [!Ref ClusterScaleDownPolicy]
      ComparisonOperator: LessThanThreshold

# Cluster level AutoScaling Rules End here

Outputs:
  AutoScalingService:
    Value: !Ref AutoScalingService