def __init__(self, scope: cdk.Stack, construct_id: str, **kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
# Create the VPC resource.
self._vpc = Vpc(self, "MyVPC", cidr="10.10.0.0/16")
# Create a Security Group within the VPC that is used to allow
# management traffic from designated jump hosts.
self._sg = SecurityGroup(
self,
"MySG",
vpc=self._vpc,
allow_all_outbound=False,
description="Management traffic from jump boxes",
security_group_name="jumpbox-mgmt-traffic")
# Add ingress rules to the Security Group for the jump host
# 10.255.0.10 to TCP/22 and TCP/3389.
self._sg.add_ingress_rule(peer=Peer.ipv4("10.255.0.10/32"),
connection=Port(
protocol=Protocol.TCP,
string_representation="host1",
from_port=22,
to_port=22))
self._sg.add_ingress_rule(peer=Peer.ipv4("10.255.0.10/32"),
connection=Port(
protocol=Protocol.TCP,
string_representation="host1",
from_port=3389,
to_port=3389))
class CdkNowIGetIt(cdk.Construct):
def __init__(self, scope: cdk.Stack, construct_id: str, vpc_cidr: str,
jump_host: str, mgmt_ports: list, subnet_len: int,
**kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
# args:
# - vpc_cidr (str): The CIDR range for the VPC.
# - jump_host (str): An optional IP address for the jump host. If this
# is not specified, te Security Group will not be
# created.
# - mgmt_ports (list): A list of TCP ports which the jump host is
# allowed to connect to.
# - subnet_len (int): The prefix length for subnet CIDR addresses.
# Create the VPC resource. The VPC does not have an internet gateway,
# or NAT gateway. Subnets are created in 2 zones.
subnets = [
SubnetConfiguration(name="MyVPC-Private",
subnet_type=SubnetType.ISOLATED,
cidr_mask=subnet_len)
]
self._vpc = Vpc(self,
"MyVPC",
cidr=vpc_cidr,
max_azs=2,
nat_gateways=None,
subnet_configuration=subnets)
# Security Group only created if the jump host parameter was
# specified.
if jump_host is not None and len(jump_host) > 0:
self.create_sg(jump_host, mgmt_ports)
def create_sg(self, jump_host, mgmt_ports):
# Create a Security Group within the VPC that is used to allow
# management traffic from designated jump hosts.
self._sg = SecurityGroup(
self,
"MySG",
vpc=self._vpc,
allow_all_outbound=False,
description="Management traffic from jump boxes",
security_group_name="jumpbox-mgmt-traffic")
# Add ingress rules to the Security Group for the jump host
# TCP/22 and TCP/3389.
for port in mgmt_ports:
self._sg.add_ingress_rule(peer=Peer.ipv4(jump_host),
connection=Port(
protocol=Protocol.TCP,
string_representation="jump",
from_port=int(port),
to_port=int(port)))
def _add_default_rules(self, sg: ec2.SecurityGroup, vpc: ec2.Vpc):
""" Adds default rules to given security group.
TODO: should create a high level construct to hide those rules.
TODO: need to review those rules because we probably do not need all of them.
Current default rules are:
- All IMCP, TCP, and UDP from within the given VPC.
- All TCP traffic from MetService network.
"""
# Allow traffic coming from internal MetService network
# TODO: we should probably restrict this rule to port 22 (SSH)
sg.add_ingress_rule(
peer=ec2.Peer.ipv4(self.config["metservice_cidr"]),
connection=ec2.Port.all_tcp(),
description="Allow connections from MetService network",
)
sg.add_ingress_rule(
peer=ec2.Peer.ipv4(vpc.vpc_cidr_block),
connection=ec2.Port.all_tcp(),
description="Allow all TCP traffic from within the VPC",
)
sg.add_ingress_rule(
peer=ec2.Peer.ipv4(vpc.vpc_cidr_block),
connection=ec2.Port.all_udp(),
description="Allow all UDP traffic from within the VPC ",
)
sg.add_ingress_rule(
peer=ec2.Peer.ipv4(vpc.vpc_cidr_block),
connection=ec2.Port.all_icmp(),
description="Allow all ICMP traffic from within the VPC ",
)
def _set_common_ingress_rules(
primary: ec2.SecurityGroup,
secondary: ec2.SecurityGroup) -> ec2.SecurityGroup:
primary.add_ingress_rule(primary, ec2.Port.tcp_range(0, 65535))
primary.add_ingress_rule(primary, ec2.Port.udp_range(0, 65535))
primary.add_ingress_rule(primary, ec2.Port.icmp_type(-1))
primary.add_ingress_rule(secondary, ec2.Port.tcp_range(0, 65535))
primary.add_ingress_rule(secondary, ec2.Port.udp_range(0, 65535))
primary.add_ingress_rule(secondary, ec2.Port.icmp_type(-1))
return primary
def __init__(self, scope: core.Construct, id_: str,
num_of_azs: int) -> None:
super().__init__(scope, id_)
self.audit_vpc = Vpc(
self,
id_,
max_azs=num_of_azs,
subnet_configuration=[
#Currently IOT, AppConfig & Cloudmap are not accessable via VPC endpoint, so we use NAT GW access them
SubnetConfiguration(name=PRIVATE_SUBNET_GROUP,
subnet_type=SubnetType.PRIVATE,
cidr_mask=24),
SubnetConfiguration(name=PUBLIC_NAT_GWS_SUBNET_GROUP,
subnet_type=SubnetType.PUBLIC,
cidr_mask=24)
],
gateway_endpoints={
'S3':
GatewayVpcEndpointOptions(
service=GatewayVpcEndpointAwsService.S3,
subnets=[
SubnetSelection(subnet_group_name=PRIVATE_SUBNET_GROUP)
]),
'DynamoDb':
GatewayVpcEndpointOptions(
service=GatewayVpcEndpointAwsService.DYNAMODB,
subnets=[
SubnetSelection(subnet_group_name=PRIVATE_SUBNET_GROUP)
]),
},
enable_dns_support=True, # For the ElasticSearch Public Domain
enable_dns_hostnames=True)
self.audit_vpc.add_interface_endpoint(
'SsmVpcEndpoint',
service=InterfaceVpcEndpointAwsService.SSM,
subnets=SubnetSelection(one_per_az=True))
self.audit_vpc.add_interface_endpoint(
'SqsVpcEndpoint',
service=InterfaceVpcEndpointAwsService.SQS,
subnets=SubnetSelection(one_per_az=True))
self.audit_vpc.add_interface_endpoint(
'Ec2VpcEndpoint',
service=InterfaceVpcEndpointAwsService.EC2,
subnets=SubnetSelection(one_per_az=True))
self.audit_vpc.add_interface_endpoint(
'LambdaVpcEndpoint',
service=InterfaceVpcEndpointAwsService.LAMBDA_,
subnets=SubnetSelection(one_per_az=True))
self.lambdas_sg = SecurityGroup(self,
id='LambdaSg',
vpc=self.audit_vpc,
security_group_name='Audit-Lambda')
def __init__(self, scope: core.Construct, id: str, vpc: Vpc, **kwargs):
super(ClusterSecurityGroupStack, self).__init__(scope, id, **kwargs)
self.__scope = scope
self.manager_security_group = SecurityGroup(scope=self,
id="manager-sg",
vpc=vpc,
allow_all_outbound=True)
self.master_security_group = SecurityGroup(scope=self,
id="master-sg",
vpc=vpc,
allow_all_outbound=True)
self.worker_security_group = SecurityGroup(scope=self,
id="worker-sg",
vpc=vpc,
allow_all_outbound=True)
self.__create_rules()
def get_nfs_security_group(self, vpc):
nfs_security_group = SecurityGroup(
self._stack,
'obm_efs',
vpc=vpc,
allow_all_outbound=True,
)
self._tag_it(nfs_security_group)
return nfs_security_group
def add_asg_fleet(self, scope: BaseApp, cluster: Cluster,
fleet) -> List[AutoScalingGroup]:
created_fleets: List[AutoScalingGroup] = []
node_labels = fleet.get('nodeLabels', {})
node_labels["fleetName"] = fleet.get('name')
node_labels_as_str = ','.join(map('='.join, node_labels.items()))
# Source of tweaks: https://kubedex.com/90-days-of-aws-eks-in-production/
kubelet_extra_args = ' '.join([
# Add node labels
f'--node-labels {node_labels_as_str}'
if len(node_labels_as_str) else '',
# Capture resource reservation for kubernetes system daemons like the kubelet, container runtime,
# node problem detector, etc.
'--kube-reserved cpu=250m,memory=1Gi,ephemeral-storage=1Gi',
# Capture resources for vital system functions, such as sshd, udev.
'--system-reserved cpu=250m,memory=0.2Gi,ephemeral-storage=1Gi',
# Start evicting pods from this node once these thresholds are crossed.
'--eviction-hard memory.available<0.2Gi,nodefs.available<10%',
])
cluster_sg = SecurityGroup.from_security_group_id(
self,
'eks-cluster-sg',
security_group_id=cluster.cluster_security_group_id)
asg_tags = {
"k8s.io/cluster-autoscaler/enabled": "true",
f"k8s.io/cluster-autoscaler/{cluster.cluster_name}": "owned",
}
# For correctly autoscaling the cluster we need our autoscaling groups to not span across AZs
# to avoid the AZ Rebalance, hence we create an ASG per subnet
for counter, subnet in enumerate(cluster.vpc.private_subnets):
asg: AutoScalingGroup = cluster.add_capacity(
id=scope.prefixed_str(f'{fleet.get("name")}-{counter}'),
instance_type=InstanceType(fleet.get('instanceType')),
min_capacity=fleet.get('autoscaling', {}).get('minInstances'),
max_capacity=fleet.get('autoscaling', {}).get('maxInstances'),
bootstrap_options=BootstrapOptions(
kubelet_extra_args=kubelet_extra_args, ),
spot_price=str(fleet.get('spotPrice'))
if fleet.get('spotPrice') else None,
vpc_subnets=SubnetSelection(subnets=[subnet]),
)
created_fleets.append(asg)
self._add_userdata_production_tweaks(asg)
for key, value in asg_tags.items():
Tag.add(asg, key, value)
return created_fleets
def create_sg(self, jump_host, mgmt_ports):
# Create a Security Group within the VPC that is used to allow
# management traffic from designated jump hosts.
self._sg = SecurityGroup(
self,
"MySG",
vpc=self._vpc,
allow_all_outbound=False,
description="Management traffic from jump boxes",
security_group_name="jumpbox-mgmt-traffic")
# Add ingress rules to the Security Group
for port in mgmt_ports:
self._sg.add_ingress_rule(peer=Peer.ipv4(jump_host),
connection=Port(
protocol=Protocol.TCP,
string_representation="jump",
from_port=int(port),
to_port=int(port)))
def get_file_system(scope: Construct) -> FileSystem:
config = get_volume_config()
stack_name = config.stack_name
security_group = SecurityGroup.from_security_group_id(
scope,
'nfs_security_group',
security_group_id=Fn.import_value(stack_name + 'SecurityGroupId'))
return FileSystem.from_file_system_attributes(
scope,
'filesystem',
file_system_id=Fn.import_value(stack_name + 'FileSystemId'),
security_group=security_group)
class CdkNowIGetIt(cdk.Construct):
def __init__(self, scope: cdk.Stack, construct_id: str, vpc_cidr: str,
jump_host: str, mgmt_ports: list, **kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
# args:
# - vpc_cidr (str): The CIDR range for the VPC.
# - jump_host (str): An optional IP address for the jump host. If this
# is not specified, the Security Group will not be
# created.
# - mgmt_ports (list): A list of TCP ports which the jump host is
# allowed to connect to.
# Create the VPC resource with the given CIDR range.
self._vpc = Vpc(self, "MyVPC", cidr=vpc_cidr)
# Security Group only created if the jump host parameter was
# specified.
if jump_host is not None and len(jump_host) > 0:
self.create_sg(jump_host, mgmt_ports)
def create_sg(self, jump_host, mgmt_ports):
# Create a Security Group within the VPC that is used to allow
# management traffic from designated jump hosts.
self._sg = SecurityGroup(
self,
"MySG",
vpc=self._vpc,
allow_all_outbound=False,
description="Management traffic from jump boxes",
security_group_name="jumpbox-mgmt-traffic")
# Add ingress rules to the Security Group
for port in mgmt_ports:
self._sg.add_ingress_rule(peer=Peer.ipv4(jump_host),
connection=Port(
protocol=Protocol.TCP,
string_representation="jump",
from_port=int(port),
to_port=int(port)))
def __init__(self, scope: Construct, id: str, vpc: IVpc, **kwargs) -> None:
super().__init__(scope, id, **kwargs)
Tags.of(self).add("Stack", "Common-Ecs")
self._cluster = Cluster(
self,
"Cluster",
vpc=vpc,
)
asg = AutoScalingGroup(
self,
"ClusterASG",
vpc=vpc,
instance_type=InstanceType("t3a.small"),
machine_image=EcsOptimizedImage.amazon_linux2(),
min_capacity=4,
)
self._cluster.add_auto_scaling_group(asg)
# Create a SecurityGroup that the NLB can use to allow traffic from
# NLB to us. This avoids a cyclic dependency.
self.security_group = SecurityGroup(
self,
"SecurityGroup",
vpc=vpc,
allow_all_outbound=False,
)
# Only use "source_security_group" to check if flows come from ECS.
# Do not use it to allow traffic in ECS; use "security_group" for
# that.
assert isinstance(asg.node.children[0], SecurityGroup)
self.source_security_group = asg.node.children[0]
# We could also make an additional security-group and add that to
# the ASG, but it keeps adding up. This makes it a tiny bit
# easier to get an overview what traffic is allowed from the
# console on AWS.
asg.node.children[0].add_ingress_rule(
peer=self.security_group,
connection=Port.tcp_range(32768, 65535),
description="NLB-self to target",
)
asg.node.children[0].add_ingress_rule(
peer=self.security_group,
connection=Port.udp_range(32768, 65535),
description="NLB-self to target (UDP)",
)
def get_web_security_group(self, vpc):
security_group = SecurityGroup(
self._stack,
'obm_web',
vpc=vpc,
allow_all_outbound=True,
)
for port_number in [SSH_PORT, HTTP_PORT, HTTPS_PORT]:
port = Port(from_port=port_number,
to_port=port_number,
protocol=Protocol.TCP,
string_representation=f"Port {port_number}")
security_group.add_ingress_rule(peer=Peer.any_ipv4(),
connection=port)
security_group.add_ingress_rule(peer=Peer.any_ipv6(),
connection=port)
self._tag_it(security_group)
return security_group
def __init__(self, scope: core.Construct, id: str, deploy_env: str,
vpc: aws_ec2.Vpc, db_redis_stack: RdsElasticacheEfsStack,
config: dict, **kwargs) -> None:
super().__init__(scope, id, **kwargs)
self.config = config
self.deploy_env = deploy_env
self.db_port = DB_PORT
# cannot map volumes to Fargate task defs yet - so this is done via Boto3 since CDK does not
# support it yet: https://github.com/aws/containers-roadmap/issues/825
#self.efs_file_system_id = db_redis_stack.efs_file_system_id
cluster_name = get_cluster_name(deploy_env)
self.cluster = ecs.Cluster(self,
cluster_name,
cluster_name=cluster_name,
vpc=vpc)
pwd_secret = ecs.Secret.from_ssm_parameter(
StringParameter.from_secure_string_parameter_attributes(
self,
f"dbpwd-{deploy_env}",
version=1,
parameter_name="postgres_pwd"))
self.secrets = {"POSTGRES_PASSWORD": pwd_secret}
environment = {
"EXECUTOR":
"Celery",
"POSTGRES_HOST":
db_redis_stack.db_host,
"POSTGRES_PORT":
str(self.db_port),
"POSTGRES_DB":
"airflow",
"POSTGRES_USER":
self.config["dbadmin"],
"REDIS_HOST":
db_redis_stack.redis_host,
"VISIBILITY_TIMEOUT":
str(self.config["celery_broker_visibility_timeout"])
}
image_asset = DockerImageAsset(self,
"AirflowImage",
directory="build",
repository_name=config["ecr_repo_name"])
self.image = ecs.ContainerImage.from_docker_image_asset(image_asset)
# web server - this initializes the db so must happen first
self.web_service = self.airflow_web_service(environment)
# https://github.com/aws/aws-cdk/issues/1654
self.web_service_sg().connections.allow_to_default_port(
db_redis_stack.postgres_db, 'allow PG')
redis_port_info = Port(protocol=Protocol.TCP,
string_representation="allow to redis",
from_port=REDIS_PORT,
to_port=REDIS_PORT)
worker_port_info = Port(protocol=Protocol.TCP,
string_representation="allow to worker",
from_port=AIRFLOW_WORKER_PORT,
to_port=AIRFLOW_WORKER_PORT)
redis_sg = SecurityGroup.from_security_group_id(
self,
id=f"Redis-SG-{deploy_env}",
security_group_id=db_redis_stack.redis.vpc_security_group_ids[0])
bastion_sg = db_redis_stack.bastion.connections.security_groups[0]
self.web_service_sg().connections.allow_to(redis_sg, redis_port_info,
'allow Redis')
self.web_service_sg().connections.allow_to_default_port(
db_redis_stack.efs_file_system)
# scheduler
self.scheduler_service = self.create_scheduler_ecs_service(environment)
# worker
self.worker_service = self.worker_service(environment)
self.scheduler_sg().connections.allow_to_default_port(
db_redis_stack.postgres_db, 'allow PG')
self.scheduler_sg().connections.allow_to(redis_sg, redis_port_info,
'allow Redis')
self.scheduler_sg().connections.allow_to_default_port(
db_redis_stack.efs_file_system)
self.worker_sg().connections.allow_to_default_port(
db_redis_stack.postgres_db, 'allow PG')
self.worker_sg().connections.allow_to(redis_sg, redis_port_info,
'allow Redis')
self.worker_sg().connections.allow_to_default_port(
db_redis_stack.efs_file_system)
# When you start an airflow worker, airflow starts a tiny web server
# subprocess to serve the workers local log files to the airflow main
# web server, who then builds pages and sends them to users. This defines
# the port on which the logs are served. It needs to be unused, and open
# visible from the main web server to connect into the workers.
self.web_service_sg().connections.allow_to(self.worker_sg(),
worker_port_info,
'web service to worker')
def __init__(self,
scope: core.Construct,
id: str,
stack_name: str,
task_definition_cpu: int,
task_definition_memory_limit_mib: int,
docker_image_name: str,
container_port: int,
desired_container_count: int,
private_subnets: Sequence[aws_ec2.Subnet] = None,
public_subnets: Sequence[aws_ec2.Subnet] = None,
private_security_group: aws_ec2.SecurityGroup = None,
public_security_group: aws_ec2.SecurityGroup = None,
vpc: aws_ec2.Vpc = None,
fargate_cluster: aws_ecs.Cluster = None,
authorizer_lambda_arn: str = None,
authorizer_lambda_role_arn: str = None,
**kwargs):
super().__init__(scope, id, **kwargs)
# Role
self.role = aws_iam.Role(
self,
'Role',
assumed_by=aws_iam.ServicePrincipal(service='ecs.amazonaws.com'),
managed_policies=[
aws_iam.ManagedPolicy.from_aws_managed_policy_name(
managed_policy_name=
'service-role/AmazonECSTaskExecutionRolePolicy')
],
inline_policies={
id:
aws_iam.PolicyDocument(statements=[
aws_iam.PolicyStatement(
effect=aws_iam.Effect.ALLOW,
actions=[
'kms:Encrypt',
'kms:Decrypt',
'kms:ReEncrypt*',
'kms:GenerateDataKey*',
'kms:DescribeKey',
'ec2:CreateNetworkInterface',
'ec2:DescribeNetworkInterfaces',
'ec2:DeleteNetworkInterface',
# Remaining actions from https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/quickref-ecs.html
'elasticloadbalancing:DeregisterInstancesFromLoadBalancer',
'elasticloadbalancing:DeregisterTargets',
'elasticloadbalancing:Describe*',
'elasticloadbalancing:RegisterInstancesWithLoadBalancer',
'elasticloadbalancing:RegisterTargets',
'ec2:Describe*',
'ec2:AuthorizeSecurityGroupIngress'
],
resources=['*'])
])
})
self.role.assume_role_policy.add_statements(
aws_iam.PolicyStatement(
actions=['sts:AssumeRole'],
principals=[
aws_iam.ServicePrincipal(service='ecs-tasks.amazonaws.com')
]))
# Set Defaults if parameters are None
if vpc is None:
vpc = aws_ec2.Vpc(self, 'Vpc')
if private_subnets is None:
private_subnets = vpc.private_subnets
if public_subnets is None:
public_subnets = vpc.public_subnets
if public_security_group is None:
public_security_group = aws_ec2.SecurityGroup(
self, 'PublicSecurityGroup', vpc=vpc, allow_all_outbound=True)
# Allow inbound HTTP traffic
public_security_group.add_ingress_rule(
peer=aws_ec2.Peer.ipv4(cidr_ip='0.0.0.0/0'),
connection=aws_ec2.Port.tcp(port=80))
# Allow inbound HTTPS traffic
public_security_group.add_ingress_rule(
peer=aws_ec2.Peer.ipv4(cidr_ip='0.0.0.0/0'),
connection=aws_ec2.Port.tcp(port=443))
if private_security_group is None:
private_security_group = aws_ec2.SecurityGroup(
self, 'PrivateSecurityGroup', vpc=vpc, allow_all_outbound=True)
public_subnet_cidr_blocks = Utils.get_subnet_cidr_blocks(
public_subnets)
# Create an ingress rule for each of the NLB's subnet's CIDR ranges and add the rules to the ECS service's
# security group. This will allow requests from the NLB to go into the ECS service. This allow inbound
# traffic from public subnets.
for cidr_block in public_subnet_cidr_blocks:
private_security_group.add_ingress_rule(
peer=aws_ec2.Peer.ipv4(cidr_ip=cidr_block),
connection=aws_ec2.Port.tcp(port=container_port))
if fargate_cluster is None:
fargate_cluster = aws_ecs.Cluster(
self,
'FargateCluster',
)
task_def = aws_ecs.FargateTaskDefinition(
self,
'TaskDefinition',
cpu=task_definition_cpu,
memory_limit_mib=task_definition_memory_limit_mib,
task_role=self.role,
execution_role=self.role)
container = aws_ecs.ContainerDefinition(
self,
'Container',
image=aws_ecs.ContainerImage.from_registry(name=docker_image_name),
task_definition=task_def,
logging=aws_ecs.AwsLogDriver(stream_prefix='/ecs'))
container.add_port_mappings(
aws_ecs.PortMapping(container_port=container_port,
protocol=aws_ec2.Protocol.TCP))
ecs_service = aws_ecs.FargateService(
self,
'FargateService',
cluster=fargate_cluster,
task_definition=task_def,
vpc_subnets=aws_ec2.SubnetSelection(subnets=private_subnets),
security_group=private_security_group,
desired_count=desired_container_count)
target_group = aws_elasticloadbalancingv2.NetworkTargetGroup(
self,
'TargetGroup',
port=80, # Health check occurs over HTTP
health_check=aws_elasticloadbalancingv2.HealthCheck(
protocol=aws_elasticloadbalancingv2.Protocol.TCP),
targets=[ecs_service],
vpc=vpc)
nlb = aws_elasticloadbalancingv2.NetworkLoadBalancer(
self,
'NetworkLoadBalancer',
vpc=vpc,
internet_facing=False,
vpc_subnets=aws_ec2.SubnetSelection(subnets=public_subnets),
)
nlb.add_listener(
id='Listener',
port=80, # HTTP listener
default_target_groups=[target_group])
# nlb.log_access_logs( # todo: add this later when you have time to research the correct bucket policy.
# bucket=aws_s3.Bucket(
# self, 'LoadBalancerLogBucket',
# bucket_name='load-balancer-logs',
# public_read_access=False,
# block_public_access=aws_s3.BlockPublicAccess(
# block_public_policy=True,
# restrict_public_buckets=True
# )
# )
# )
# Dependencies
ecs_service.node.add_dependency(nlb)
# API Gateway
rest_api = aws_apigateway.RestApi(self, stack_name)
resource = rest_api.root.add_resource(
path_part='{proxy+}',
default_method_options=aws_apigateway.MethodOptions(
request_parameters={'method.request.path.proxy': True}))
token_authorizer = None
if authorizer_lambda_arn and authorizer_lambda_role_arn:
token_authorizer = aws_apigateway.TokenAuthorizer( #todo: make this a parameter?
self,
'JwtTokenAuthorizer',
results_cache_ttl=core.Duration.minutes(5),
identity_source='method.request.header.Authorization',
assume_role=aws_iam.Role.from_role_arn(
self,
'AuthorizerLambdaInvokationRole',
role_arn=authorizer_lambda_role_arn),
handler=aws_lambda.Function.from_function_arn(
self,
'AuthorizerLambda',
function_arn=authorizer_lambda_arn))
resource.add_method(
http_method='ANY',
authorization_type=aws_apigateway.AuthorizationType.CUSTOM,
authorizer=token_authorizer,
integration=aws_apigateway.HttpIntegration(
url=f'http://{nlb.load_balancer_dns_name}/{{proxy}}',
http_method='ANY',
proxy=True,
options=aws_apigateway.IntegrationOptions(
request_parameters={
'integration.request.path.proxy':
'method.request.path.proxy'
},
connection_type=aws_apigateway.ConnectionType.VPC_LINK,
vpc_link=aws_apigateway.VpcLink(
self,
'VpcLink',
description=
f'API Gateway VPC Link to internal NLB for {stack_name}',
vpc_link_name=stack_name,
targets=[nlb]))))
def __init__(self, scope: cdk.Construct, construct_id: str, config, vpc: IVpc, **kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
self.neo4j_user_secret = Secret(self,'secretsmanager-secret-neo4j-user',
secret_name=NEO4J_USER_SECRET_NAME
)
neo4j_server_instance_role = Role(self,'iam-role-neo4j-server-instance',
assumed_by=ServicePrincipal('ec2.amazonaws.com'),
managed_policies=[
ManagedPolicy.from_aws_managed_policy_name('AmazonSSMManagedInstanceCore'), # Use SSM Session Manager rather than straight ssh
ManagedPolicy.from_aws_managed_policy_name('CloudWatchAgentServerPolicy')
],
inline_policies={
"work-with-tags": PolicyDocument(
statements=[
PolicyStatement(
actions=[
'ec2:CreateTags',
'ec2:Describe*',
'elasticloadbalancing:Describe*',
'cloudwatch:ListMetrics',
'cloudwatch:GetMetricStatistics',
'cloudwatch:Describe*',
'autoscaling:Describe*',
],
resources=["*"]
)
]
),
"access-neo4j-user-secret": PolicyDocument(
statements=[
PolicyStatement(
actions=['secretsmanager:GetSecretValue'],
resources=[self.neo4j_user_secret.secret_arn]
)
]
)
}
)
instance_security_group = SecurityGroup(self, "ec2-sg-neo4j-server-instance",
description="Altimeter Neo4j Server Instance",
vpc=vpc
)
instance_security_group.add_ingress_rule(Peer.ipv4("0.0.0.0/0"), Port.tcp(7687), 'Bolt from ANYWHERE') # TESTING
# instance_security_group.add_ingress_rule(Peer.ipv4("0.0.0.0/0"), Port.tcp(7473), 'Bolt from ANYWHERE') # TESTING
# Prepare userdata script
with open("./resources/neo4j-server-instance-userdata.sh", "r") as f:
userdata_content = f.read()
userdata_content = userdata_content.replace("[[NEO4J_USER_SECRET_NAME]]",NEO4J_USER_SECRET_NAME)
user_data = UserData.for_linux()
user_data.add_commands(userdata_content)
instance_type = InstanceType.of(InstanceClass.BURSTABLE2, InstanceSize.MEDIUM)
self.instance = Instance(self, 'ec2-instance-neo4j-server-instance',
instance_name="instance-altimeter--neo4j-community-standalone-server",
machine_image=MachineImage.generic_linux(
ami_map={
"eu-west-1": "ami-00c8631d384ad7c53"
}
),
instance_type=instance_type,
role=neo4j_server_instance_role,
vpc=vpc,
# vpc_subnets=SubnetSelection(subnets=vpc.select_subnets(subnet_group_name='Private').subnets),
vpc_subnets=SubnetSelection(subnets=vpc.select_subnets(subnet_group_name='Public').subnets),
security_group=instance_security_group,
user_data=user_data,
block_devices=[
BlockDevice(
device_name="/dev/sda1",
volume=BlockDeviceVolume.ebs(
volume_size=10,
volume_type=EbsDeviceVolumeType.GP2,
encrypted=True, # Just encrypt
delete_on_termination=True
)
),
BlockDevice(
device_name="/dev/sdb",
volume=BlockDeviceVolume.ebs(
volume_size=20, # TODO: Check size requirements
volume_type=EbsDeviceVolumeType.GP2,
encrypted=True,
delete_on_termination=True # ASSUMPTION: NO 'primary' data, only altimeter results.
)
)
]
)
cdk.Tags.of(self.instance).add("neo4j_mode", "SINGLE")
cdk.Tags.of(self.instance).add("dbms_mode", "SINGLE")
def __init__(self, scope: core.Construct, id: str, **kwargs) -> None:
super().__init__(scope, id, **kwargs)
attribute_tagged_group = Group(self, "Flexible Tagged")
access_project = core.CfnTag(key="access-project", value="elysian")
access_team = core.CfnTag(key="access-team", value="webdev")
access_cost_center = core.CfnTag(key="cost-center", value="2600")
flexible_boundary_policy = CfnManagedPolicy(
self,
"FlexiblePermissionBoundary",
policy_document=json.loads(flexible_policy_permission_boundary),
)
CfnUser(
self,
"Developer",
tags=[access_project, access_team, access_cost_center],
groups=[attribute_tagged_group.group_name],
permissions_boundary=flexible_boundary_policy.ref,
)
# Add AWS managed policy for EC2 Read Only access for the console.
attribute_tagged_group.add_managed_policy(
ManagedPolicy.from_aws_managed_policy_name(
managed_policy_name="AmazonEC2ReadOnlyAccess"
)
)
# Import a json policy and create CloudFormation Managed Policy
CfnManagedPolicy(
self,
"FlexibleAttributePolicy",
policy_document=json.loads(full_attribute_based_policy),
groups=[attribute_tagged_group.group_name],
)
vpc = Vpc.from_lookup(self, "AttributeTaggedVPC", is_default=True)
instance_type = InstanceType("t2.micro")
ami = MachineImage.latest_amazon_linux()
blocked_instance = Instance(
self,
"Blocked Instance",
machine_image=ami,
instance_type=instance_type,
vpc=vpc,
)
# Re-use the AMI from t
image_id = blocked_instance.instance.image_id
# Can only add tags to CfnInstance as of cdk v1.31
valid_instance = CfnInstance(
self,
"Valid Instance",
image_id=image_id,
instance_type="t2.micro",
tags=[access_project, access_team, access_cost_center],
)
# Empty group as it's not need to complete our tests.
test_security_group = SecurityGroup(
self, "EmptySecurityGroup", vpc=vpc)
core.CfnOutput(
self,
"BlockedInstance",
value=blocked_instance.instance_id,
export_name="elysian-blocked-instance",
)
core.CfnOutput(
self,
"ValidInstance",
value=valid_instance.ref,
export_name="elysian-valid-instance",
)
core.CfnOutput(
self,
"TestSecurityGroup",
value=test_security_group.security_group_id,
export_name="test-elysian-sg",
)
core.CfnOutput(
self, "DefaultAMI", value=image_id, export_name="default-elysian-ami"
)
def __init__(self, scope: core.Construct, id: str, **kwargs) -> None:
super().__init__(scope, id, **kwargs)
username_tagged = Group(self, "Username Tagged")
developer = User(self, "Developer")
developer.add_to_group(username_tagged)
# Add AWS managed policy for EC2 Read Only access for the console.
username_tagged.add_managed_policy(
ManagedPolicy.from_aws_managed_policy_name(
managed_policy_name="AmazonEC2ReadOnlyAccess"
)
)
# Import a json policy and create CloudFormation Managed Policy
CfnManagedPolicy(
self,
"UserTaggedPolicy",
policy_document=json.loads(username_based_policy),
groups=[username_tagged.group_name],
)
vpc = Vpc.from_lookup(self, "UsernameTaggedVPC", is_default=True)
instance_type = InstanceType("t2.micro")
ami = MachineImage.latest_amazon_linux()
blocked_instance = Instance(
self,
"Blocked Instance",
machine_image=ami,
instance_type=instance_type,
vpc=vpc,
)
# Re-use the AMI from t
image_id = blocked_instance.instance.image_id
# Can only add tags to CfnInstance as of 1.31
dev_username_tag = core.CfnTag(
key="username", value=developer.user_name)
valid_instance = CfnInstance(
self,
"Valid Instance",
image_id=image_id,
instance_type="t2.micro",
tags=[dev_username_tag],
)
# Empty group as it's not need to complete our tests.
test_security_group = SecurityGroup(
self, "EmptySecurityGroup", vpc=vpc)
core.CfnOutput(
self,
"BlockedInstance",
value=blocked_instance.instance_id,
export_name="username-blocked-instance",
)
core.CfnOutput(
self,
"ValidInstance",
value=valid_instance.ref,
export_name="username-valid-instance",
)
core.CfnOutput(
self,
"TestSecurityGroup",
value=test_security_group.security_group_id,
export_name="test-username-sg",
)
core.CfnOutput(
self, "DefaultAMI", value=image_id, export_name="default-username-ami"
)
def __init__(self, scope: core.Construct, construct_id: str,
**kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
role = iam.Role(
scope=self,
id='AwsCustomResourceRole',
assumed_by=iam.ServicePrincipal('lambda.amazonaws.com'))
role.add_to_policy(
iam.PolicyStatement(actions=['iam:PassRole'], resources=['*']))
my_custom_resource = cr.AwsCustomResource(
scope=self,
id='MyAwsCustomResource',
role=role,
policy=cr.AwsCustomResourcePolicy.from_sdk_calls(resources=['*']),
on_create=cr.AwsSdkCall(
action='listBuckets',
service='s3',
physical_resource_id=cr.PhysicalResourceId.of('BucketsList'),
))
vpc = VPCConstruct(self, id_='test-vpc', num_of_azs=2)
security_group = SecurityGroup(
self,
id='test-security-group',
vpc=vpc,
security_group_name='test-security-group')
security_group.add_ingress_rule(connection=Port.tcp(443),
peer=vpc.lambdas_sg)
domain = es.Domain(
scope=self,
id='Domain',
version=es.ElasticsearchVersion.V7_9,
domain_name="es-domain-name",
enable_version_upgrade=False,
enforce_https=True,
fine_grained_access_control=None,
node_to_node_encryption=True,
tls_security_policy=es.TLSSecurityPolicy.TLS_1_0,
logging=es.LoggingOptions(
app_log_enabled=True,
slow_index_log_enabled=True,
slow_search_log_enabled=True,
app_log_group=LogGroup(
scope=self,
id="app-log-group",
log_group_name=f'/aws/aes/domains/esdomain/app-log-group',
removal_policy=core.RemovalPolicy.DESTROY),
slow_index_log_group=LogGroup(
scope=self,
id="slow-index-log-group",
log_group_name=
f'/aws/aes/domains/esdomain/slow-index-log-group',
removal_policy=core.RemovalPolicy.DESTROY),
slow_search_log_group=LogGroup(
scope=self,
id="slow-search-log-group",
log_group_name=
f'/aws/aes/domains/esdomain/slow-search-log-group',
removal_policy=core.RemovalPolicy.DESTROY)),
removal_policy=core.RemovalPolicy.DESTROY,
zone_awareness=es.ZoneAwarenessConfig(availability_zone_count=2,
enabled=True),
vpc_options=es.VpcOptions(
security_groups=[security_group],
subnets=vpc.audit_vpc.select_subnets(
subnet_group_name=PRIVATE_SUBNET_GROUP).subnets))
def __init__(self, scope: core.Construct, construct_id: str,
**kwargs) -> None:
super().__init__(scope, construct_id, **kwargs)
# Create two VPCs - one to host our private website, the other to act as a client
website_vpc = Vpc(
self,
"WEBSITEVPC",
cidr="10.0.0.0/16",
)
client_vpc = Vpc(
self,
"ClientVPC",
cidr="10.1.0.0/16",
)
# Create a bastion host in the client API which will act like our client workstation
bastion = BastionHostLinux(
self,
"WEBClient",
vpc=client_vpc,
instance_name='my-bastion',
instance_type=InstanceType('t3.micro'),
machine_image=AmazonLinuxImage(),
subnet_selection=SubnetSelection(subnet_type=SubnetType.PRIVATE),
security_group=SecurityGroup(
scope=self,
id='bastion-sg',
security_group_name='bastion-sg',
description=
'Security group for the bastion, no inbound open because we should access'
' to the bastion via AWS SSM',
vpc=client_vpc,
allow_all_outbound=True))
# Set up a VPC peering connection between client and API VPCs, and adjust
# the routing table to allow connections back and forth
VpcPeeringHelper(self, 'Peering', website_vpc, client_vpc)
# Create VPC endpoints for API gateway
vpc_endpoint = InterfaceVpcEndpoint(
self,
'APIGWVpcEndpoint',
vpc=website_vpc,
service=InterfaceVpcEndpointAwsService.APIGATEWAY,
private_dns_enabled=True,
)
vpc_endpoint.connections.allow_from(bastion, Port.tcp(443))
endpoint_id = vpc_endpoint.vpc_endpoint_id
api_policy = iam.PolicyDocument(statements=[
iam.PolicyStatement(principals=[iam.AnyPrincipal()],
actions=['execute-api:Invoke'],
resources=['execute-api:/*'],
effect=iam.Effect.DENY,
conditions={
"StringNotEquals": {
"aws:SourceVpce": endpoint_id
}
}),
iam.PolicyStatement(principals=[iam.AnyPrincipal()],
actions=['execute-api:Invoke'],
resources=['execute-api:/*'],
effect=iam.Effect.ALLOW)
])
# Create an s3 bucket to hold the content
content_bucket = s3.Bucket(self,
"ContentBucket",
removal_policy=core.RemovalPolicy.DESTROY)
# Upload our static content to the bucket
s3dep.BucketDeployment(self,
"DeployWithInvalidation",
sources=[s3dep.Source.asset('website')],
destination_bucket=content_bucket)
# Create a private API GW in the API VPC
api = apigw.RestApi(self,
'PrivateS3Api',
endpoint_configuration=apigw.EndpointConfiguration(
types=[apigw.EndpointType.PRIVATE],
vpc_endpoints=[vpc_endpoint]),
policy=api_policy)
# Create a role to allow API GW to access our S3 bucket contents
role = iam.Role(
self,
"Role",
assumed_by=iam.ServicePrincipal("apigateway.amazonaws.com"))
role.add_to_policy(
iam.PolicyStatement(effect=iam.Effect.ALLOW,
resources=[
content_bucket.bucket_arn,
content_bucket.bucket_arn + '/*'
],
actions=["s3:Get*"]))
# Create a proxy resource that captures all non-root resource requests
resource = api.root.add_resource("{proxy+}")
# Create an integration with S3
resource_integration = apigw.Integration(
type=apigw.IntegrationType.AWS,
integration_http_method='GET',
options=apigw.IntegrationOptions(
request_parameters=
{ # map the proxy parameter so we can pass the request path
"integration.request.path.proxy":
"method.request.path.proxy"
},
integration_responses=[
apigw.IntegrationResponse(
status_code='200',
response_parameters=
{ # map the content type of the S3 object back to the HTTP response
"method.response.header.Content-Type":
"integration.response.header.Content-Type"
})
],
credentials_role=role),
# reference the bucket content we want to retrieve
uri='arn:aws:apigateway:eu-west-1:s3:path/%s/{proxy}' %
(content_bucket.bucket_name))
# handle the GET request and map it to our new integration
resource.add_method(
"GET",
resource_integration,
method_responses=[
apigw.MethodResponse(status_code='200',
response_parameters={
"method.response.header.Content-Type":
False
})
],
request_parameters={"method.request.path.proxy": True})
# Handle requests to the root of our site
# Create another integration with S3 - this time with no proxy parameter
resource_integration = apigw.Integration(
type=apigw.IntegrationType.AWS,
integration_http_method='GET',
options=apigw.IntegrationOptions(
integration_responses=[
apigw.IntegrationResponse(
status_code='200',
response_parameters=
{ # map the content type of the S3 object back to the HTTP response
"method.response.header.Content-Type":
"integration.response.header.Content-Type"
})
],
credentials_role=role),
# reference the bucket content we want to retrieve
uri='arn:aws:apigateway:eu-west-1:s3:path/%s/index.html' %
(content_bucket.bucket_name))
# handle the GET request and map it to our new integration
api.root.add_method("GET",
resource_integration,
method_responses=[
apigw.MethodResponse(
status_code='200',
response_parameters={
"method.response.header.Content-Type":
False
})
])
def __init__(
self,
scope: Construct,
id: str,
*,
vpc: IVpc,
cluster: ICluster,
service: IEc2Service,
ecs_security_group: SecurityGroup,
deployment: Deployment,
**kwargs,
) -> None:
super().__init__(scope, id, **kwargs)
Tags.of(self).add("Application", self.application_name)
Tags.of(self).add("Deployment", deployment.value)
security_group = SecurityGroup(
self,
"LambdaSG",
vpc=vpc,
)
lambda_func = Function(
self,
"ReloadLambda",
code=Code.from_asset("./lambdas/bananas-reload"),
handler="index.lambda_handler",
runtime=Runtime.PYTHON_3_8,
timeout=Duration.seconds(120),
environment={
"CLUSTER": cluster.cluster_arn,
"SERVICE": service.service_arn,
},
vpc=vpc,
security_groups=[security_group, ecs_security_group],
reserved_concurrent_executions=1,
)
lambda_func.add_to_role_policy(
PolicyStatement(
actions=[
"ec2:DescribeInstances",
"ecs:DescribeContainerInstances",
"ecs:DescribeTasks",
"ecs:ListContainerInstances",
"ecs:ListServices",
"ecs:ListTagsForResource",
"ecs:ListTasks",
],
resources=[
"*",
],
)
)
policy = ManagedPolicy(self, "Policy")
policy.add_statements(
PolicyStatement(
actions=[
"lambda:InvokeFunction",
],
resources=[lambda_func.function_arn],
)
)
def __init__(
self,
scope: Construct,
id: str,
cluster: ICluster,
ecs_security_group: SecurityGroup,
ecs_source_security_group: SecurityGroup,
vpc: IVpc,
**kwargs,
) -> None:
super().__init__(scope, id, **kwargs)
global g_nlb
Tags.of(self).add("Stack", "Common-Nlb")
# TODO -- You need to do some manual actions:
# TODO -- 1) enable auto-assign IPv6 address on public subnets
# TODO -- 2) add to the Outbound rules of "Live-Common-Nlb/ASG/InstanceSecurityGroup" the destination "::/0"
self.private_zone = HostedZone.from_lookup(
self,
"PrivateZone",
domain_name="openttd.internal",
private_zone=True,
)
user_data = UserData.for_linux(shebang="#!/bin/bash -ex")
asset = Asset(self, "NLB", path="user_data/nlb/")
user_data.add_commands(
"echo 'Extracting user-data files'",
"mkdir /nlb",
"cd /nlb",
)
user_data.add_s3_download_command(
bucket=asset.bucket,
bucket_key=asset.s3_object_key,
local_file="/nlb/files.zip",
)
user_data.add_commands("unzip files.zip", )
user_data.add_commands(
"echo 'Setting up configuration'",
f"echo '{self.region}' > /etc/.region",
f"echo '{cluster.cluster_name}' > /etc/.cluster",
)
user_data.add_commands(
"echo 'Installing nginx'",
"amazon-linux-extras install epel",
"yum install nginx -y",
"cp /nlb/nginx.conf /etc/nginx/nginx.conf",
"mkdir /etc/nginx/nlb.d",
)
user_data.add_commands(
"echo 'Installing Python3'",
"yum install python3 -y",
"python3 -m venv /venv",
"/venv/bin/pip install -r /nlb/requirements.txt",
)
user_data.add_commands(
"echo 'Generating nginx configuration'",
"cd /etc/nginx/nlb.d",
"/venv/bin/python /nlb/nginx.py",
"systemctl start nginx",
)
user_data.add_commands(
"echo 'Setting up SOCKS proxy'",
"useradd pproxy",
"cp /nlb/pproxy.service /etc/systemd/system/",
"systemctl daemon-reload",
"systemctl enable pproxy.service",
"systemctl start pproxy.service",
)
asg = AutoScalingGroup(
self,
"ASG",
vpc=vpc,
instance_type=InstanceType("t3a.nano"),
machine_image=MachineImage.latest_amazon_linux(
generation=AmazonLinuxGeneration.AMAZON_LINUX_2),
min_capacity=2,
vpc_subnets=SubnetSelection(subnet_type=SubnetType.PUBLIC,
one_per_az=True),
user_data=user_data,
health_check=HealthCheck.elb(grace=Duration.seconds(0)),
)
asg.add_security_group(ecs_security_group)
asg.role.add_managed_policy(
ManagedPolicy.from_aws_managed_policy_name(
"AmazonSSMManagedInstanceCore"))
asset.grant_read(asg.role)
policy = ManagedPolicy(self, "Policy")
policy_statement = PolicyStatement(
actions=[
"ec2:DescribeInstances",
"ecs:DescribeContainerInstances",
"ecs:DescribeTasks",
"ecs:ListContainerInstances",
"ecs:ListServices",
"ecs:ListTagsForResource",
"ecs:ListTasks",
],
resources=["*"],
)
policy.add_statements(policy_statement)
asg.role.add_managed_policy(policy)
# We could also make an additional security-group and add that to
# the ASG, but it keeps adding up. This makes it a tiny bit
# easier to get an overview what traffic is allowed from the
# console on AWS.
assert isinstance(asg.node.children[0], SecurityGroup)
self.security_group = asg.node.children[0]
listener_https.add_targets(
subdomain_name=self.admin_subdomain_name,
port=80,
target=asg,
priority=2,
)
# Create a Security Group so the lambdas can access the EC2.
# This is needed to check if the EC2 instance is fully booted.
lambda_security_group = SecurityGroup(
self,
"LambdaSG",
vpc=vpc,
)
self.security_group.add_ingress_rule(
peer=lambda_security_group,
connection=Port.tcp(80),
description="Lambda to target",
)
self.security_group.add_ingress_rule(
peer=ecs_source_security_group,
connection=Port.udp(8080),
description="ECS to target",
)
self.create_ecs_lambda(
cluster=cluster,
auto_scaling_group=asg,
)
self.create_asg_lambda(
lifecycle_transition=LifecycleTransition.INSTANCE_LAUNCHING,
timeout=Duration.seconds(180),
vpc=vpc,
security_group=lambda_security_group,
auto_scaling_group=asg,
)
self.create_asg_lambda(
lifecycle_transition=LifecycleTransition.INSTANCE_TERMINATING,
timeout=Duration.seconds(30),
vpc=vpc,
security_group=lambda_security_group,
auto_scaling_group=asg,
)
# Initialize the NLB record on localhost, as we need to be able to
# reference it for other entries to work correctly.
ARecord(
self,
"ARecord",
target=RecordTarget.from_ip_addresses("127.0.0.1"),
zone=dns.get_hosted_zone(),
record_name=self.subdomain_name,
ttl=Duration.seconds(60),
)
AaaaRecord(
self,
"AAAARecord",
target=RecordTarget.from_ip_addresses("::1"),
zone=dns.get_hosted_zone(),
record_name=self.subdomain_name,
ttl=Duration.seconds(60),
)
# To make things a bit easier, also alias to staging.
self.create_alias(self, "nlb.staging")
# Create a record for the internal DNS
ARecord(
self,
"APrivateRecord",
target=RecordTarget.from_ip_addresses("127.0.0.1"),
zone=self.private_zone,
record_name=self.subdomain_name,
ttl=Duration.seconds(60),
)
if g_nlb is not None:
raise Exception("Only a single NlbStack instance can exist")
g_nlb = self