cf_kubeflow_single_instance.yaml

AWSTemplateFormatVersion: 2010-09-09
Mappings:
  RegionMap:
    eu-central-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    us-east-1:
      AMI: ami-0b93ce03dcbcb10f6
    us-gov-east-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    us-gov-west-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    us-east-2:
      AMI: ami-xxxxxxxxxxxxxxxxx
    us-west-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    us-west-2:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ca-central-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    eu-west-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    eu-west-2:
      AMI: ami-xxxxxxxxxxxxxxxxx
    eu-west-3:
      AMI: ami-xxxxxxxxxxxxxxxxx
    eu-north-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    eu-south-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-east-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-southeast-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-southeast-2:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-south-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-northeast-1:
      AMI: ami-xxxxxxxxxxxxxxxxx
    ap-northeast-2:
      AMI: ami-xxxxxxxxxxxxxxxxx
    me-south-1:
      AMI: ami-xxxxxxxxxxxxxxxxx 
Description: ""
Parameters:
  AppName:
    Description: Name of the application the cluster will serve
    Type: String
    MinLength: "3"
    AllowedPattern: '[a-zA-Z0-9\(\)\.\-/_]+'
    Default: "kubeflow-single-instance"
  KeyPair:
    Description: Amazon EC2 Key Pair used to ssh to the cluster nodes
    Type: "AWS::EC2::KeyPair::KeyName"
  InstanceTypeParameter:
    Type: String
    Default: t2.2xlarge
    AllowedValues:
      - t2.2xlarge
    Description: Select the instance type of the cluster nodes
  InstanceVolumeSize:
    Type: Number
    Default: 100
    MinValue: 100
    Description: Size of the (unencripted DeleteOnTermination) gp2 volume attatched to the instance
  KubernetesVersion:
    Type: String
    Default: 1.22
    AllowedValues:
      - 1.23
      - 1.22
      - 1.21
    Description: The Kubernetes version
  KubeflowVersion:
    Type: String
    Default: 1.6/stable
    AllowedValues:
      - 1.6/stable
      - latest/stable
      - latest/edge
    Description: The Kubeflow version to be deployed
  KubeflowDashboardUsername:
    Type: String
    MinLength: "5"
    AllowedPattern: '[a-zA-Z0-9\(\)\.\-/_@]+'
    Default: "user123@email.com"
  KubeflowDashboardPassword:
    Type: String
    NoEcho : "true"
    MinLength: "5"
    AllowedPattern: '[a-zA-Z0-9\(\)\.\-/_@]+'
    Default: "user123"
    Description: Default is user123
  VpcCIDR:
    Description: Please enter the IP range (CIDR notation) for this VPC or use Default
    Type: String
    Default: 10.192.0.0/16
  PublicSubnet1CIDR:
    Description: Please enter the IP range (CIDR notation) for the public subnet (random Availability Zone will be assigned)
    Type: String
    Default: 10.192.10.0/24
Resources:
  SecurityGroupEgress:
    Type: AWS::EC2::SecurityGroup
    Properties:
      GroupName: Kubeflow allow all egress rule
      GroupDescription: "Kubeflow instance security group without ingress rules"
      VpcId: !Ref VPC
  
  EC2LaunchTemplate:
    Type: "AWS::EC2::LaunchTemplate"
    DependsOn: SecurityGroupEgress
    Properties:
      LaunchTemplateName: !Sub "${AppName}"
      LaunchTemplateData:
        UserData:
          Fn::Base64: !Sub |
            #!/bin/bash

            apt update
            apt -y upgrade
            for snap in juju juju-wait charmcraft; do sudo snap install $snap --classic; done
            snap install microk8s --classic --channel=${KubernetesVersion}/stable
            sudo snap refresh charmcraft --channel latest/candidate
            usermod -a -G microk8s ubuntu
            mkdir /home/ubuntu/.kube
            chown -f -R ubuntu /home/ubuntu/.kube

            microk8s enable dns storage metallb:"10.64.140.43-10.64.140.49,192.168.0.105-192.168.0.111"
            sleep 120
            microk8s.kubectl wait --for=condition=available -nkube-system deployment/coredns deployment/hostpath-provisioner
            microk8s.kubectl -n kube-system rollout status ds/calico-node

            su ubuntu -c 'juju bootstrap microk8s uk8s-controller'
            su ubuntu -c 'juju add-model kubeflow'
            su ubuntu -c 'juju deploy kubeflow --channel=${KubeflowVersion} --trust'

            su ubuntu -c 'juju config dex-auth public-url=http://10.64.140.43.nip.io; juju config oidc-gatekeeper public-url=http://10.64.140.43.nip.io; juju config dex-auth static-username=${KubeflowDashboardUsername}; juju config dex-auth static-password=${KubeflowDashboardPassword}'
            sleep 720
            echo "Charmed Kubeflow deployed"

            su ubuntu -c 'juju run --unit istio-pilot/0 -- "export JUJU_DISPATCH_PATH=hooks/config-changed; ./dispatch"'
            su ubuntu -c 'juju deploy mlflow-server'
            su ubuntu -c 'juju deploy charmed-osm-mariadb-k8s mlflow-db'
            su ubuntu -c 'juju relate minio mlflow-server'
            su ubuntu -c 'juju relate istio-pilot mlflow-server'
            su ubuntu -c 'juju relate mlflow-db mlflow-server'
            su ubuntu -c 'juju relate mlflow-server admission-webhook'
            
            echo "Charmed MlFlow deployed"
        BlockDeviceMappings:
          - DeviceName: "/dev/sda1"
            Ebs:
              Encrypted: false
              DeleteOnTermination: true
              VolumeSize: !Ref InstanceVolumeSize
              VolumeType: "gp2"
        IamInstanceProfile:
          Arn: !GetAtt
            - InstanceProfile
            - Arn
        EbsOptimized: false
        KeyName: !Sub "${KeyPair}"
        DisableApiTermination: false
        ImageId: !FindInMap
          - RegionMap
          - !Ref AWS::Region
          - AMI
        InstanceType: !Sub "${InstanceTypeParameter}"
        SecurityGroupIds:
          - !Ref SecurityGroupEgress
  EC2Role:
    Type: "AWS::IAM::Role"
    Properties:
      Description: !Sub "${AppName} Role"
      RoleName: !Sub "${AppName}-EC2Role"
      AssumeRolePolicyDocument:
        Version: "2012-10-17"
        Statement:
          - Effect: Allow
            Principal:
              Service:
                - ec2.amazonaws.com
            Action:
              - "sts:AssumeRole"
      Policies:
        - PolicyName: root
          PolicyDocument:
            Version: "2012-10-17"
            Statement:
              - Action: ec2:CreateTags
                Effect: Allow
                Resource: "arn:*:ec2:*:*:instance/*"
              - Action: ec2:DescribeTags
                Effect: Allow
                Resource: "*"
  AutoScalingAutoScalingGroup:
    Type: "AWS::AutoScaling::AutoScalingGroup"
    Properties:
      AutoScalingGroupName: !Sub "${AppName}"
      LaunchTemplate:
        LaunchTemplateId: !Ref EC2LaunchTemplate
        Version: 1
      MinSize: 1
      MaxSize: 1
      DesiredCapacity: 1
      Cooldown: 300
      VPCZoneIdentifier:
        - !Ref PublicSubnet
      HealthCheckType: "EC2"
      HealthCheckGracePeriod: 300
      TerminationPolicies:
        - "NewestInstance"
      ServiceLinkedRoleARN: !Sub "arn:aws:iam::${AWS::AccountId}:role/aws-service-role/autoscaling.amazonaws.com/AWSServiceRoleForAutoScaling_${AppName}"
      NewInstancesProtectedFromScaleIn: false
    DependsOn:
      - EC2LaunchTemplate
      - AutoScalingRole
      - PublicSubnet
      - PublicSubnetRouteTableAssociation
      - SecurityGroupEgress
  AutoScalingScalingPolicy:
    Type: "AWS::AutoScaling::ScalingPolicy"
    Properties:
      AutoScalingGroupName: !Sub "${AppName}"
      PolicyType: "SimpleScaling"
      AdjustmentType: "ChangeInCapacity"
      ScalingAdjustment: 1
    DependsOn: "AutoScalingAutoScalingGroup"
  AutoScalingRole:
    Type: "AWS::IAM::ServiceLinkedRole"
    Properties:
      AWSServiceName: autoscaling.amazonaws.com
      Description: AutoScaling ServiceLinked Role
      CustomSuffix: !Sub "${AppName}"
  
  InstanceProfile:
    Type: AWS::IAM::InstanceProfile
    DependsOn: "EC2Role"
    Properties:
      InstanceProfileName: !Sub "${AppName}-InstanceProfile"
      Roles:
        - !Ref EC2Role
  
  VPC:
    Type: AWS::EC2::VPC
    Properties:
      CidrBlock: !Ref VpcCIDR
      EnableDnsSupport: true
      EnableDnsHostnames: true
      Tags:
      - Key: Name
        Value:  !Join ['', [!Ref "AWS::StackName", "-VPC" ]]
  InternetGateway:
    Type: AWS::EC2::InternetGateway
    DependsOn: VPC
  AttachGateway:
    Type: AWS::EC2::VPCGatewayAttachment
    Properties:
      VpcId: !Ref VPC
      InternetGatewayId: !Ref InternetGateway
  PublicSubnet:
    Type: AWS::EC2::Subnet
    Properties:
      VpcId: !Ref VPC
      CidrBlock: !Ref PublicSubnet1CIDR
      AvailabilityZone: !Select [ 0, !GetAZs '' ]
      MapPublicIpOnLaunch: true
      Tags:
      - Key: Name
        Value: !Sub ${AWS::StackName}-Public-A
  PublicRouteTable:
    Type: AWS::EC2::RouteTable
    Properties:
      VpcId: !Ref VPC
      Tags:
      - Key: Name
        Value: Public
  PublicRoute1:
    Type: AWS::EC2::Route
    DependsOn: AttachGateway
    Properties:
      RouteTableId: !Ref PublicRouteTable
      DestinationCidrBlock: 0.0.0.0/0
      GatewayId: !Ref InternetGateway
  PublicSubnetRouteTableAssociation:
    Type: AWS::EC2::SubnetRouteTableAssociation
    Properties:
      SubnetId: !Ref PublicSubnet
      RouteTableId: !Ref PublicRouteTable