CloudFront Origins for the CDK CloudFront Library

This library contains convenience methods for defining origins for a CloudFront distribution. You can use this library to create origins from S3 buckets, Elastic Load Balancing v2 load balancers, or any other domain name.

S3 Bucket

An S3 bucket can be used as an origin. An S3 bucket origin can either be configured using a standard S3 bucket or using a S3 bucket that’s configured as a website endpoint (see AWS docs for Using an S3 Bucket).

Note: S3Origin has been deprecated. Use S3BucketOrigin for standard S3 origins and S3StaticWebsiteOrigin for static website S3 origins.

Standard S3 Bucket

To set up an origin using a standard S3 bucket, use the S3BucketOrigin class. The bucket is handled as a bucket origin and CloudFront’s redirect and error handling will be used. It is recommended to use S3BucketOrigin.withOriginAccessControl() to configure OAC for your origin.

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket))
)

Note: When you use CloudFront OAC with Amazon S3 bucket origins, you must set Amazon S3 Object Ownership to Bucket owner enforced (the default for new Amazon S3 buckets). If you require ACLs, use the Bucket owner preferred setting to maintain control over objects uploaded via CloudFront.

S3 Bucket Configured as a Website Endpoint

To set up an origin using an S3 bucket configured as a website endpoint, use the S3StaticWebsiteOrigin class. When the bucket is configured as a website endpoint, the bucket is treated as an HTTP origin, and the distribution can use built-in S3 redirects and S3 custom error pages.

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.S3StaticWebsiteOrigin(my_bucket))
)

Restricting access to a standard S3 Origin

CloudFront provides two ways to send authenticated requests to a standard Amazon S3 origin:

origin access control (OAC) and
origin access identity (OAI)

OAI is considered legacy due to limited functionality and regional limitations, whereas OAC is recommended because it supports all Amazon S3 buckets in all AWS Regions, Amazon S3 server-side encryption with AWS KMS (SSE-KMS), and dynamic requests (PUT and DELETE) to Amazon S3. Additionally, OAC provides stronger security posture with short term credentials, and more frequent credential rotations as compared to OAI. OAI and OAC can be used in conjunction with a bucket that is not public to require that your users access your content using CloudFront URLs and not S3 URLs directly.

See AWS docs on Restricting access to an Amazon S3 Origin for more details.

Note: OAC and OAI can only be used with an regular S3 bucket origin (not a bucket configured as a website endpoint).

The S3BucketOrigin class supports creating a standard S3 origin with OAC, OAI, and no access control (using your bucket access settings) via the withOriginAccessControl(), withOriginAccessIdentity(), and withBucketDefaults() methods respectively.

Setting up a new origin access control (OAC)

Setup a standard S3 origin with origin access control as follows:

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket)
    )
)

When creating a standard S3 origin using origins.S3BucketOrigin.withOriginAccessControl(), an Origin Access Control resource is automatically created with the origin type set to s3 and signing behavior set to always.

You can grant read, write or delete access to the OAC using the originAccessLevels property:

my_bucket = s3.Bucket(self, "myBucket")
s3_origin = origins.S3BucketOrigin.with_origin_access_control(my_bucket,
    origin_access_levels=[cloudfront.AccessLevel.READ, cloudfront.AccessLevel.WRITE, cloudfront.AccessLevel.DELETE]
)

You can also pass in a custom S3 origin access control:

my_bucket = s3.Bucket(self, "myBucket")
oac = cloudfront.S3OriginAccessControl(self, "MyOAC",
    signing=cloudfront.Signing.SIGV4_NO_OVERRIDE
)
s3_origin = origins.S3BucketOrigin.with_origin_access_control(my_bucket,
    origin_access_control=oac
)
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=s3_origin
    )
)

An existing S3 origin access control can be imported using the fromOriginAccessControlId method:

imported_oAC = cloudfront.S3OriginAccessControl.from_origin_access_control_id(self, "myImportedOAC", "ABC123ABC123AB")

Note: When you use OAC with S3 bucket origins, the bucket’s object ownership must be either set to Bucket owner enforced (default for new S3 buckets) or Bucket owner preferred (only if you require ACLs).

Setting up OAC with a SSE-KMS encrypted S3 origin

If the objects in the S3 bucket origin are encrypted using server-side encryption with AWS Key Management Service (SSE-KMS), the OAC must have permission to use the KMS key.

Setting up a standard S3 origin using S3BucketOrigin.withOriginAccessControl() will automatically add the statement to the KMS key policy to give the OAC permission to use the KMS key.

import aws_cdk.aws_kms as kms


my_kms_key = kms.Key(self, "myKMSKey")
my_bucket = s3.Bucket(self, "mySSEKMSEncryptedBucket",
    encryption=s3.BucketEncryption.KMS,
    encryption_key=my_kms_key,
    object_ownership=s3.ObjectOwnership.BUCKET_OWNER_ENFORCED
)
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket)
    )
)

Scoping down the key policy

I saw this warning message during synth time. What do I do?

To avoid a circular dependency between the KMS key, Bucket, and Distribution during the initial deployment, a wildcard is used in the Key policy condition to match all Distribution IDs.
After deploying once, it is strongly recommended to further scope down the policy for best security practices by following the guidance in the "Using OAC for a SSE-KMS encrypted S3 origin" section in the module README.

If the S3 bucket has an encryptionKey defined, S3BucketOrigin.withOriginAccessControl() will automatically add the following policy statement to the KMS key policy to allow CloudFront read-only access (unless otherwise specified in the originAccessLevels property).

{
    "Statement": {
        "Effect": "Allow",
        "Principal": {
            "Service": "cloudfront.amazonaws.com"
        },
        "Action": "kms:Decrypt",
        "Resource": "*",
        "Condition": {
            "ArnLike": {
                "AWS:SourceArn": "arn:aws:cloudfront::<account ID>:distribution/*"
            }
        }
    }
}

This policy uses a wildcard to match all distribution IDs in the account instead of referencing the specific distribution ID to resolve the circular dependency. The policy statement is not as scoped down as the example in the AWS CloudFront docs (see SSE-KMS section).

After you have deployed the Distribution, you should follow these steps to only grant permissions to the specific distribution according to AWS best practices:

Step 1. Copy the key policy

Step 2. Use an escape hatch to update the policy statement condition so that

  "Condition": {
      "ArnLike": {
          "AWS:SourceArn": "arn:aws:cloudfront::<account ID>:distribution/*"
      }
  }

…becomes…

  "Condition": {
      "StringEquals": {
          "AWS:SourceArn": "arn:aws:cloudfront::111122223333:distribution/<CloudFront distribution ID>"
      }
  }

Note the change of condition operator from ArnLike to StringEquals in addition to replacing the wildcard (*) with the distribution ID.

To set the key policy using an escape hatch:

import aws_cdk.aws_kms as kms


kms_key = kms.Key(self, "myKMSKey")
my_bucket = s3.Bucket(self, "mySSEKMSEncryptedBucket",
    encryption=s3.BucketEncryption.KMS,
    encryption_key=kms_key,
    object_ownership=s3.ObjectOwnership.BUCKET_OWNER_ENFORCED
)
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket)
    )
)

# Add the following to scope down the key policy
scoped_down_key_policy = {
    "Version": "2012-10-17",
    "Statement": [{
        "Effect": "Allow",
        "Principal": {
            "AWS": "arn:aws:iam::111122223333:root"
        },
        "Action": "kms:*",
        "Resource": "*"
    }, {
        "Effect": "Allow",
        "Principal": {
            "Service": "cloudfront.amazonaws.com"
        },
        "Action": ["kms:Decrypt", "kms:Encrypt", "kms:GenerateDataKey*"
        ],
        "Resource": "*",
        "Condition": {
            "StringEquals": {
                "AWS:SourceArn": "arn:aws:cloudfront::111122223333:distribution/<CloudFront distribution ID>"
            }
        }
    }
    ]
}
cfn_key = (kms_key.node.default_child)
cfn_key.key_policy = scoped_down_key_policy

Step 3. Deploy the stack

Tip: Run cdk diff before deploying to verify the changes to your stack.

Step 4. Verify your final key policy includes the following statement after deploying:

{
    "Effect": "Allow",
    "Principal": {
        "Service": [
            "cloudfront.amazonaws.com"
        ]
     },
    "Action": [
        "kms:Decrypt",
        "kms:Encrypt",
        "kms:GenerateDataKey*"
    ],
    "Resource": "*",
    "Condition": {
            "StringEquals": {
                "AWS:SourceArn": "arn:aws:cloudfront::111122223333:distribution/<CloudFront distribution ID>"
            }
        }
}

Updating imported key policies

If you are using an imported KMS key to encrypt your S3 bucket and want to use OAC, you will need to update the key policy manually to allow CloudFront to use the key. Like most imported resources, CDK apps cannot modify the configuration of imported keys.

After deploying the distribution, add the following policy statement to your key policy to allow CloudFront OAC to access your KMS key for SSE-KMS:

{
    "Sid": "AllowCloudFrontServicePrincipalSSE-KMS",
    "Effect": "Allow",
    "Principal": {
        "Service": [
            "cloudfront.amazonaws.com"
        ]
     },
    "Action": [
        "kms:Decrypt",
        "kms:Encrypt",
        "kms:GenerateDataKey*"
    ],
    "Resource": "*",
    "Condition": {
            "StringEquals": {
                "AWS:SourceArn": "arn:aws:cloudfront::111122223333:distribution/<CloudFront distribution ID>"
            }
        }
}

See CloudFront docs on SSE-KMS for more details.

Setting up OAC with imported S3 buckets

If you are using an imported bucket for your S3 Origin and want to use OAC, you will need to update the S3 bucket policy manually to allow the OAC to access the S3 origin. Like most imported resources, CDK apps cannot modify the configuration of imported buckets.

After deploying the distribution, add the following policy statement to your S3 bucket to allow CloudFront read-only access (or additional S3 permissions as required):

{
    "Version": "2012-10-17",
    "Statement": {
        "Effect": "Allow",
        "Principal": {
            "Service": "cloudfront.amazonaws.com"
        },
        "Action": "s3:GetObject",
        "Resource": "arn:aws:s3:::<S3 bucket name>/*",
        "Condition": {
            "StringEquals": {
                "AWS:SourceArn": "arn:aws:cloudfront::111122223333:distribution/<CloudFront distribution ID>"
            }
        }
    }
}

See CloudFront docs on Giving the origin access control permission to access the S3 bucket for more details.

Note: If your bucket previously used OAI, you will need to manually remove the policy statement that gives the OAI access to your bucket after setting up OAC.

Setting up an OAI (legacy)

Setup an S3 origin with origin access identity (legacy) as follows:

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.S3BucketOrigin.with_origin_access_identity(my_bucket)
    )
)

You can also pass in a custom S3 origin access identity:

my_bucket = s3.Bucket(self, "myBucket")
my_oai = cloudfront.OriginAccessIdentity(self, "myOAI",
    comment="My custom OAI"
)
s3_origin = origins.S3BucketOrigin.with_origin_access_identity(my_bucket,
    origin_access_identity=my_oai
)
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=s3_origin
    )
)

Setting up OAI with imported S3 buckets (legacy)

If you are using an imported bucket for your S3 Origin and want to use OAI, you will need to update the S3 bucket policy manually to allow the OAI to access the S3 origin. Like most imported resources, CDK apps cannot modify the configuration of imported buckets.

Add the following policy statement to your S3 bucket to allow the OAI read access:

{
    "Version": "2012-10-17",
    "Id": "PolicyForCloudFrontPrivateContent",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "AWS": "arn:aws:iam::cloudfront:user/CloudFront Origin Access Identity <origin access identity ID>"
            },
            "Action": "s3:GetObject",
            "Resource": "arn:aws:s3:::<S3 bucket name>/*"
        }
    ]
}

See AWS docs on Giving an origin access identity permission to read files in the Amazon S3 bucket for more details.

Setting up a S3 origin with no origin access control

To setup a standard S3 origin with no access control (no OAI nor OAC), use origins.S3BucketOrigin.withBucketDefaults():

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.S3BucketOrigin.with_bucket_defaults(my_bucket)
    )
)

Migrating from OAI to OAC

If you are currently using OAI for your S3 origin and wish to migrate to OAC, replace the S3Origin construct (deprecated) with S3BucketOrigin.withOriginAccessControl() which automatically creates and sets up an OAC for you.

Existing setup using OAI and S3Origin:

my_bucket = s3.Bucket(self, "myBucket")
s3_origin = origins.S3Origin(my_bucket)
distribution = cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=s3_origin)
)

Step 1:

To ensure CloudFront doesn’t lose access to the bucket during the transition, add a statement to bucket policy to grant OAC access to the S3 origin. Deploy the stack. If you are okay with downtime during the transition, you can skip this step.

Tip: Run cdk diff before deploying to verify the changes to your stack.

import aws_cdk as cdk
import aws_cdk.aws_iam as iam


stack = Stack()
my_bucket = s3.Bucket(self, "myBucket")
s3_origin = origins.S3Origin(my_bucket)
distribution = cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=s3_origin)
)

# Construct the bucket policy statement
distribution_arn = stack.format_arn(
    service="cloudfront",
    region="",
    resource="distribution",
    resource_name=distribution.distribution_id,
    arn_format=cdk.ArnFormat.SLASH_RESOURCE_NAME
)

cloudfront_sP = iam.ServicePrincipal("cloudfront.amazonaws.com")

oac_bucket_policy_statement = iam.PolicyStatement(
    effect=iam.Effect.ALLOW,
    principals=[cloudfront_sP],
    actions=["s3:GetObject"],
    resources=[my_bucket.arn_for_objects("*")],
    conditions={
        "StringEquals": {
            "AWS:SourceArn": distribution_arn
        }
    }
)

# Add statement to bucket policy
my_bucket.add_to_resource_policy(oac_bucket_policy_statement)

The following changes will take place:

The bucket policy will be modified to grant the CloudFront distribution access. At this point the bucket policy allows both an OAI and an OAC to access the S3 origin.

Step 2:

Replace S3Origin with S3BucketOrigin.withOriginAccessControl(), which creates an OAC and attaches it to the distribution. You can remove the code from Step 1 which updated the bucket policy, as S3BucketOrigin.withOriginAccessControl() updates the bucket policy automatically with the same statement when defined in the Distribution (no net difference).

Run cdk diff before deploying to verify the changes to your stack.

bucket = s3.Bucket(self, "Bucket")
s3_origin = origins.S3BucketOrigin.with_origin_access_control(bucket)
distribution = cloudfront.Distribution(self, "Distribution",
    default_behavior=cloudfront.BehaviorOptions(origin=s3_origin)
)

The following changes will take place:

A AWS::CloudFront::OriginAccessControl resource will be created.
The Origin property of the AWS::CloudFront::Distribution will set OriginAccessControlId to the OAC ID after it is created. It will also set S3OriginConfig to {"OriginAccessIdentity": ""}, which deletes the origin access identity from the existing distribution.
The AWS::CloudFront::CloudFrontOriginAccessIdentity resource will be deleted.

Will migrating from OAI to OAC cause any resource replacement?

No, following the migration steps does not cause any replacement of the existing AWS::CloudFront::Distribution, AWS::S3::Bucket nor AWS::S3::BucketPolicy resources. It will modify the bucket policy, create a AWS::CloudFront::OriginAccessControl resource, and delete the existing AWS::CloudFront::CloudFrontOriginAccessIdentity.

Will migrating from OAI to OAC have any availability implications for my application?

Updates to bucket policies are eventually consistent. Therefore, removing OAI permissions and setting up OAC in the same CloudFormation stack deployment is not recommended as it may cause downtime where CloudFront loses access to the bucket. Following the steps outlined above lowers the risk of downtime as the bucket policy is updated to have both OAI and OAC permissions, then in a subsequent deployment, the OAI permissions are removed.

For more information, see Migrating from origin access identity (OAI) to origin access control (OAC).

Adding Custom Headers

You can configure CloudFront to add custom headers to the requests that it sends to your origin. These custom headers enable you to send and gather information from your origin that you don’t get with typical viewer requests. These headers can even be customized for each origin. CloudFront supports custom headers for both for custom and Amazon S3 origins.

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket,
        custom_headers={
            "Foo": "bar"
        }
    ))
)

ELBv2 Load Balancer

An Elastic Load Balancing (ELB) v2 load balancer may be used as an origin. In order for a load balancer to serve as an origin, it must be publicly accessible (internetFacing is true). Both Application and Network load balancers are supported.

import aws_cdk.aws_ec2 as ec2
import aws_cdk.aws_elasticloadbalancingv2 as elbv2

# vpc: ec2.Vpc

# Create an application load balancer in a VPC. 'internetFacing' must be 'true'
# for CloudFront to access the load balancer and use it as an origin.
lb = elbv2.ApplicationLoadBalancer(self, "LB",
    vpc=vpc,
    internet_facing=True
)
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.LoadBalancerV2Origin(lb))
)

The origin can also be customized to respond on different ports, have different connection properties, etc.

import aws_cdk.aws_elasticloadbalancingv2 as elbv2

# load_balancer: elbv2.ApplicationLoadBalancer

origin = origins.LoadBalancerV2Origin(load_balancer,
    connection_attempts=3,
    connection_timeout=Duration.seconds(5),
    read_timeout=Duration.seconds(45),
    keepalive_timeout=Duration.seconds(45),
    protocol_policy=cloudfront.OriginProtocolPolicy.MATCH_VIEWER
)

Note that the readTimeout and keepaliveTimeout properties can extend their values over 60 seconds only if a limit increase request for CloudFront origin response timeout quota has been approved in the target account; otherwise, values over 60 seconds will produce an error at deploy time. Consider that this value is still limited to a maximum value of 180 seconds, which is a hard limit for that quota.

From an HTTP endpoint

Origins can also be created from any other HTTP endpoint, given the domain name, and optionally, other origin properties.

cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.HttpOrigin("www.example.com"))
)

See the documentation of aws-cdk-lib/aws-cloudfront for more information.

Failover Origins (Origin Groups)

You can set up CloudFront with origin failover for scenarios that require high availability. To get started, you create an origin group with two origins: a primary and a secondary. If the primary origin is unavailable, or returns specific HTTP response status codes that indicate a failure, CloudFront automatically switches to the secondary origin. You achieve that behavior in the CDK using the OriginGroup class:

my_bucket = s3.Bucket(self, "myBucket")
cloudfront.Distribution(self, "myDist",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.OriginGroup(
            primary_origin=origins.S3BucketOrigin.with_origin_access_control(my_bucket),
            fallback_origin=origins.HttpOrigin("www.example.com"),
            # optional, defaults to: 500, 502, 503 and 504
            fallback_status_codes=[404]
        )
    )
)

From an API Gateway REST API

Origins can be created from an API Gateway REST API. It is recommended to use a regional API in this case. The origin path will automatically be set as the stage name.

# api: apigateway.RestApi

cloudfront.Distribution(self, "Distribution",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.RestApiOrigin(api))
)

If you want to use a different origin path, you can specify it in the originPath property.

# api: apigateway.RestApi

cloudfront.Distribution(self, "Distribution",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.RestApiOrigin(api, origin_path="/custom-origin-path"))
)

From a Lambda Function URL

Lambda Function URLs enable direct invocation of Lambda functions via HTTP(S), without intermediaries. They can be set as CloudFront origins for streamlined function execution behind a CDN, leveraging caching and custom domains.

import aws_cdk.aws_lambda as lambda_

# fn: lambda.Function

fn_url = fn.add_function_url(auth_type=lambda_.FunctionUrlAuthType.NONE)

cloudfront.Distribution(self, "Distribution",
    default_behavior=cloudfront.BehaviorOptions(origin=origins.FunctionUrlOrigin(fn_url))
)

Lambda Function URL with Origin Access Control (OAC)

You can configure the Lambda Function URL with Origin Access Control (OAC) for enhanced security. When using OAC with Signing SIGV4_ALWAYS, it is recommended to set the Lambda Function URL authType to AWS_IAM to ensure proper authorization.

import aws_cdk.aws_lambda as lambda_
# fn: lambda.Function


fn_url = fn.add_function_url(
    auth_type=lambda_.FunctionUrlAuthType.AWS_IAM
)

cloudfront.Distribution(self, "MyDistribution",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.FunctionUrlOrigin.with_origin_access_control(fn_url)
    )
)

If you want to explicitly add OAC for more customized access control, you can use the originAccessControl option as shown below.

import aws_cdk.aws_lambda as lambda_
# fn: lambda.Function


fn_url = fn.add_function_url(
    auth_type=lambda_.FunctionUrlAuthType.AWS_IAM
)

# Define a custom OAC
oac = cloudfront.FunctionUrlOriginAccessControl(self, "MyOAC",
    origin_access_control_name="CustomLambdaOAC",
    signing=cloudfront.Signing.SIGV4_ALWAYS
)

# Set up Lambda Function URL with OAC in CloudFront Distribution
cloudfront.Distribution(self, "MyDistribution",
    default_behavior=cloudfront.BehaviorOptions(
        origin=origins.FunctionUrlOrigin.with_origin_access_control(fn_url,
            origin_access_control=oac
        )
    )
)