メンテナンスウィンドウからのターゲットグループ引き渡し
SSMドキュメントでは入力パラメータ―InstanceID
を指定するものがよくありますが、
メンテナンスウィンドウでは何もしないと空扱いになり動きません。
メンテナンスウィンドウのターゲットで指定したインスタンスは、
入力パラメータ―に{{ TARGET_ID }}
を指定することではじめて引き渡されます。
必要なIAM権限が多い
consoleから手で実行する場合と異なり、メンテナンスウィンドウはAWS上で自動的に実行されます
そのため思ったより権限が必要です。
AWS公式のSSMドキュメントを実行するときですら、公式手順に加えて適切なポリシーをつける必要があります。
https://docs.aws.amazon.com/ja_jp/systems-manager/latest/userguide/sysman-maintenance-permissions.html
Lambdaのデプロイ
公式ドキュメントにはCloudformationでLambdaをデプロイし利用後時に破壊するものが多いですが、
インスタンスごとに何度もデプロイと破壊されるので無駄が多いと感じます。
おまけ: AutoScalingグループから台数をキープしつつ1台外し、外したインスタンスを終了するSSMドキュメント
AWS公式のAWS-ASGEnterStandbyドキュメントはShouldDecrement\": true
で決め打ちのため、
中身をコピーして作りました。
{
"schemaVersion": "0.3",
"assumeRole": "{{AutomationAssumeRole}}",
"description": "ASGから1台外し、終了する。ASGから外すときにASGが1台起動する仕組み",
"parameters": {
"InstanceId": {
"type": "String",
"description": "(Required) ID of EC2 Instance to change standby state for within ASG"
},
"LambdaRoleArn": {
"default": "",
"type": "String",
"description": "(Optional) The ARN of the role that allows Lambda created by Automation to perform the actions on your behalf. If not specified a transient role will be created to execute the Lambda function."
},
"AutomationAssumeRole": {
"default": "",
"type": "String",
"description": "(Optional) The ARN of the role that allows Automation to perform the actions on your behalf."
}
},
"mainSteps": [
{
"action": "aws:createStack",
"inputs": {
"StackName": "asg-state-change-lambda-cfn-stack-{{automation:EXECUTION_ID}}",
"Parameters": [
{
"ParameterValue": "asg-state-change-lambda-{{automation:EXECUTION_ID}}",
"ParameterKey": "FunctionName"
},
{
"ParameterValue": "{{LambdaRoleArn}}",
"ParameterKey": "LambdaRoleArn"
}
],
"Capabilities": [
"CAPABILITY_IAM"
],
"TemplateBody": "AWSTemplateFormatVersion: '2010-09-09'\nConditions:\n IsVerbose:\n Fn::Equals:\n - {Ref: Verbose}\n - 'true'\n IsVerboseAndLambdaRoleNotSpecified:\n Fn::And:\n - {Condition: LambdaAssumeRoleNotSpecified}\n - {Condition: IsVerbose}\n LambdaAssumeRoleNotSpecified:\n Fn::Or:\n - Fn::Equals:\n - {Ref: LambdaRoleArn}\n - ''\n - Fn::Equals:\n - {Ref: LambdaRoleArn}\n - undefined\nDescription: Automation stack for ASG Change Standby state documents\nParameters:\n FunctionName: {Description: What to name the deployed lambda function, Type: String}\n LambdaRoleArn: {Default: '', Description: 'Assume role used by the lambda function.\n If not specified this template will create a temporary role to be used by the\n lambda created in this template.\n\n ', Type: String}\n Verbose:\n AllowedValues: ['true', 'false']\n Default: 'true'\n Description: 'Verbose setting\n\n '\n Type: String\nResources:\n ChangeASGStateLambda:\n Properties:\n Code: {ZipFile: \"import logging\\n\\nimport boto3\\n\\n\\ndef handler(event, context):\\n\\\n \\t\\\"\\\"\\\"\\n\\tChanges the state of an instance in an autoscaling group. The\\\n \\ IAM role running this lambda requires the following\\n\\tpermissions:\\n\\t\\\n {\\n\\t \\\"Effect\\\": \\\"Allow\\\",\\n\\t \\\"Action\\\": [\\n\\t\\t\\\"autoscaling:EnterStandby\\\"\\\n ,\\n\\t\\t\\\"autoscaling:ExitStandby\\\",\\n\\t\\t\\\"autoscaling:DescribeAutoScalingInstances\\n\\\n \\t ],\\n\\t \\\"Resource\\\": \\\"*\\\"\\n\\t}\\n\\t:param event: Defined fields:\\n\\t\\\n \\t{\\n\\t\\t \\\"State\\\": \\\"EnterStandby|ExitStandby\\\",\\n\\t\\t \\\"InstanceId\\\"\\\n : \\\"i-1234567890\\\",\\n\\t\\t \\\"ASGName\\\": \\\"MyASGName\\\",\\n\\t\\t \\\"ShouldDecrement\\\"\\\n : true|false\\n\\t\\t}\\n\\tThe ShouldDecrement field is only used for EnterStandby\\\n \\ and ignored otherwise\\n\\t\\\"\\\"\\\"\\n\\tas_client = boto3.client('autoscaling')\\n\\\n \\t# The state to transition to. Options are EnterStandby and ExitStandby\\n\\\n \\tstate = event.get('State')\\n\\tinstance_id = event.get('InstanceId')\\n\\t\\\n decrement = event.get('ShouldDecrement', False)\\n\\n\\tassert state in {'EnterStandby',\\\n \\ 'ExitStandby'}, 'Invalid state provided'\\n\\tassert instance_id is not\\\n \\ None, 'InstanceId must be specified'\\n\\n\\tinstances = as_client.describe_auto_scaling_instances(InstanceIds=[instance_id])\\n\\\n \\tif len(instances.get(\\\"AutoScalingInstances\\\", [])) > 0:\\n\\t\\tasg_name\\\n \\ = instances[\\\"AutoScalingInstances\\\"][0][\\\"AutoScalingGroupName\\\"]\\n\\t\\\n \\tif state == 'EnterStandby':\\n\\t\\t\\tprint \\\"Enter Standby: {} {}\\\".format(instance_id,\\\n \\ asg_name)\\n\\t\\t\\tas_client.enter_standby(InstanceIds=[instance_id],\\n\\t\\\n \\t\\t\\t\\t\\t\\t\\t\\tAutoScalingGroupName=asg_name,\\n\\t\\t\\t\\t\\t\\t\\t\\t\\tShouldDecrementDesiredCapacity=decrement)\\n\\\n \\t\\telse:\\n\\t\\t\\tprint \\\"Exit Standby: {} {}\\\".format(instance_id, asg_name)\\n\\\n \\t\\t\\tas_client.exit_standby(InstanceIds=[instance_id], AutoScalingGroupName=asg_name)\\n\"}\n FunctionName: {Ref: FunctionName}\n Handler: index.handler\n Role:\n Fn::If:\n - LambdaAssumeRoleNotSpecified\n - Fn::GetAtt: [LambdaRole, Arn]\n - {Ref: LambdaRoleArn}\n Runtime: python2.7\n Type: AWS::Lambda::Function\n LambdaLogPolicy:\n Condition: IsVerboseAndLambdaRoleNotSpecified\n Properties:\n PolicyDocument:\n Statement:\n Action: ['log:CreateLogStream', 'log:PutLogEvents', 'log:CreateLogGroup']\n Effect: Allow\n Resource: {'Fn::Sub': 'arn:aws:logs:${AWS::Region}:${AWS::AccountId}:*'}\n Version: '2012-10-17'\n PolicyName: lambda-log-access\n Roles:\n - {Ref: LambdaRole}\n Type: AWS::IAM::Policy\n LambdaRole:\n Condition: LambdaAssumeRoleNotSpecified\n Properties:\n AssumeRolePolicyDocument:\n Statement:\n - Action: ['sts:AssumeRole']\n Effect: Allow\n Principal:\n Service: [lambda.amazonaws.com]\n Version: '2012-10-17'\n Path: /\n Policies:\n - PolicyDocument:\n Statement:\n Action: ['autoscaling:EnterStandby', 'autoscaling:ExitStandby', 'autoscaling:DescribeAutoScalingInstances']\n Effect: Allow\n Resource: '*'\n Version: '2012-10-17'\n PolicyName: asg-access\n Type: AWS::IAM::Role\n"
},
"maxAttempts": 1,
"name": "deployChangeStateLambda",
"onFailure": "Abort"
},
{
"action": "aws:invokeLambdaFunction",
"inputs": {
"FunctionName": "asg-state-change-lambda-{{automation:EXECUTION_ID}}",
"Payload": "{\"InstanceId\": \"{{InstanceId}}\", \"State\": \"EnterStandby\", \"ShouldDecrement\": false}"
},
"maxAttempts": 1,
"name": "changeState",
"onFailure": "Abort"
},
{
"action": "aws:deleteStack",
"inputs": {
"StackName": "asg-state-change-lambda-cfn-stack-{{automation:EXECUTION_ID}}"
},
"maxAttempts": 1,
"name": "deleteChangeStateLambda",
"onFailure": "Abort"
},
{
"name": "TerminateEC2Instance",
"action": "aws:executeAutomation",
"maxAttempts": 1,
"timeoutSeconds": 600,
"onFailure": "Abort",
"inputs": {
"DocumentName": "AWS-TerminateEC2Instance",
"RuntimeParameters": {
"InstanceId": [
"{{InstanceId}}"
],
"AutomationAssumeRole": [
"{{AutomationAssumeRole}}"
]
}
}
},
]