I’m using an S3 website endpoint as the origin of my CloudFront distribution. Why am I getting 403 Access Denied errors?

Last updated: 2020-09-25

I'm using an Amazon Simple Storage Service (Amazon S3) bucket as the origin of my Amazon CloudFront distribution. I'm using the S3 static website endpoint as the origin domain name. Why is CloudFront returning 403 Access Denied errors from Amazon S3?

Short description

To troubleshoot Access Denied errors, determine if your distribution’s origin domain name is an S3 website endpoint or an S3 REST API endpoint. Follow these steps to determine the endpoint type:

  1. Open the CloudFront console.
  2. Choose your CloudFront distribution, and then choose Distribution Settings.
  3. Choose the Origins and Origin Groups tab.
  4. Review the domain name under Origin Domain Name and Path, and then determine the endpoint type based on the format of the domain name.

REST API endpoints use this format:

DOC-EXAMPLE-BUCKET.s3.amazonaws.com

Note: Be sure to follow the rules for naming Amazon S3 buckets.

Website endpoints use this format:

DOC-EXAMPLE-BUCKET.s3-website-us-east-1.amazonaws.com

If your distribution is using a REST API endpoint, see I'm using an S3 REST API endpoint as the origin of my CloudFront distribution. Why am I getting 403 Access Denied errors?

If your distribution is using a website endpoint, verify the following requirements to avoid Access Denied errors:

  • Objects in the bucket must be publicly accessible.
  • Objects in the bucket can't be encrypted by AWS Key Management Service (AWS KMS).
  • The bucket policy must allow access to s3:GetObject.
  • If the bucket policy grants public read access, then the AWS account that owns the bucket must also own the object.
  • The requested objects must exist in the bucket.
  • Amazon S3 Block Public Access must be disabled on the bucket.
  • If Requester Pays is enabled, then the request must include the request-payer parameter.
  • If you're using a Referer header to restrict access from CloudFront to your S3 origin, then review the custom header.

Note: If you don't want to allow public (anonymous) access to your S3 objects, then change your configuration to use the S3 REST API endpoint as the origin of your distribution. Then, configure your distribution and S3 bucket to restrict access using an origin access identity (OAI). For instructions, see Using a REST API endpoint as the origin with access restricted by an OAI in How do I use CloudFront to serve a static website hosted on Amazon S3?

Resolution

Objects in the bucket must be publicly accessible

A distribution using a website endpoint supports only publicly accessible content. To determine if an object in your S3 bucket is publicly accessible, open the object's URL in a web browser. Or, you can run a curl command on the URL.

The following is an example URL of an S3 object:

http://DOC-EXAMPLE-BUCKET.s3-website-us-east-1.amazonaws.com/index.html

If the web browser or curl command returns an Access Denied error, then the object isn't publicly accessible.

Allow public read access to the object in one of the following ways:

Objects in the bucket can't be AWS KMS-encrypted

CloudFront distributions don't support AWS KMS-encrypted objects. You must remove KMS encryption from the S3 objects that you want to serve using the distribution.

Note: Instead of using AWS KMS encryption, use AES-256 to encrypt your objects.

Use one of the following ways to check if an object in your bucket is KMS-encrypted:

To change the object's encryption settings using the Amazon S3 console, see How do I add encryption to an S3 object?

To change the object's encryption settings using the AWS CLI, first verify that the object's bucket doesn't have default encryption. If the bucket doesn't have default encryption, then run the following command to remove the object's encryption by copying the object over itself:

aws s3 cp s3://DOC-EXAMPLE-BUCKET/index.html s3://DOC-EXAMPLE-BUCKET/index.html

Warning: Copying the object over itself removes settings for storage-class and website-redirect-location. To maintain these settings in the new object, be sure to explicitly specify storage-class or website-redirect-location values in the copy request.

The bucket policy must allow access to s3:GetObject

To use a distribution with an S3 website endpoint, your bucket policy must not have a deny statement that blocks public read access to the s3:GetObject action.

Even if you have an explicit allow statement for s3:GetObject in your bucket policy, confirm that there isn't a conflicting explicit deny statement. An explicit deny statement always overrides an explicit allow statement.

Follow these steps to review your bucket policy for s3:GetObject:

1.    Open your S3 bucket from the Amazon S3 console.

2.    Choose the Permissions tab.

3.    Choose Bucket Policy.

4.    Review the bucket policy for statements with "Action": "s3:GetObject" or "Action": "s3:*".

The following example policy contains an explicit allow statement for public access to s3:GetObject. However, there's also an explicit deny statement for s3:GetObject that blocks access unless the request is from a specific Amazon Virtual Private Cloud (Amazon VPC).

{
    "Version": "2008-10-17",
    "Id": "PolicyForCloudFrontPrivateContent",
    "Statement": [
        {
            "Sid": "Allow-OAI-Access-To-Bucket",
            "Effect": "Allow",
            "Principal": {
                "AWS": "arn:aws:iam::cloudfront:user/CloudFront Origin Access Identity EAF5XXXXXXXXX"
            },
            "Action": "s3:GetObject",
            "Resource": [
                "arn:aws:s3:::DOC-EXAMPLE-BUCKET/*"
            ]
        },
        {
            "Sid": "Allow-Public-Access-To-Bucket",
            "Effect": "Allow",
            "Principal": "*",
            "Action": "s3:GetObject",
            "Resource": [
                "arn:aws:s3:::DOC-EXAMPLE-BUCKET/*"
            ]
        },
        {
            "Sid": "Access-to-specific-VPCE-only",
            "Effect": "Deny",
            "Principal": "*",
            "Action": "s3:GetObject",
            "Resource": [
                "arn:aws:s3:::DOC-EXAMPLE-BUCKET/*"
            ],
            "Condition": {
                "StringNotEquals": {
                    "aws:sourceVpce": "vpce-1a2b3c4d"
                }
            }
        }
    ]
  }

5.    Modify the bucket policy to remove or edit statements that block public read access to s3:GetObject.

Note: CloudFront caches the results of an Access Denied error for the amount of time specified in the error caching minimum TTL. The default value is one minute. After removing a deny statement from the bucket policy, you can run an invalidation on your distribution to remove the object from the cache.

If the bucket policy grants public read access, then the AWS account that owns the bucket must also own the object

For a bucket policy to allow public read access to objects, the AWS account that owns the bucket must also own the objects. A bucket or object is owned by the account of the AWS Identity and Access Management (IAM) identity that created the bucket or object.

Note: The object-ownership requirement applies to public read access granted by a bucket policy. It doesn't apply to public read access granted by the object's access control list (ACL).

Follow these steps to check if the bucket and objects have the same owner:

1.    Run this AWS CLI command to get the S3 canonical ID of the bucket owner:

aws s3api list-buckets --query Owner.ID

2.    Run this command to get the S3 canonical ID of the object owner:

Note: This example shows a single object, but you can use the list command to check several objects.

aws s3api list-objects --bucket DOC-EXAMPLE-BUCKET --prefix index.html

3.    If the canonical IDs don't match, then the bucket and object have different owners.

Note: You can also use the Amazon S3 console to check the bucket and object owners. The owners are found in the Permissions tab of the respective bucket or object.

Follow these steps to change the object's owner to the bucket owner:

1.    From the object owner's account, run this command to retrieve the ACL permissions assigned to the object:

aws s3api get-object-acl --bucket DOC-EXAMPLE-BUCKET --key object-name

2.    If the object has bucket-owner-full-control ACL permissions, then skip to step #3. If the object doesn't have bucket-owner-full-control ACL permissions, then run this command from the object owner's account:

aws s3api put-object-acl --bucket DOC-EXAMPLE-BUCKET --key object-name --acl bucket-owner-full-control

3.    From the bucket owner's account, run this command to change the owner of the object by copying the object over itself:

aws s3 cp s3://DOC-EXAMPLE-BUCKET/index.html s3://DOC-EXAMPLE-BUCKET/index.html

The requested objects must exist in the bucket

If a user doesn’t have s3:ListBucket permissions, then the user gets Access Denied errors for missing objects instead of 404 Not Found errors. Run the head-object AWS CLI command to check if an object exists in the bucket.

Note: Confirm that the object request sent to CloudFront matches the S3 object name exactly. S3 object names are case-sensitive. If the request doesn't have the correct object name, then Amazon S3 responds as though the object is missing. To identify which object CloudFront is requesting from Amazon S3, use server access logging.

If the object exists in the bucket, then the Access Denied error isn't masking a 404 Not Found error. Verify other configuration requirements to resolve the Access Denied error.

If the object isn’t in the bucket, then the Access Denied error is masking a 404 Not Found error. Resolve the issue related to the missing object.

Note: It's not a security best practice to enable public s3:ListBucket access. Enabling public s3:ListBucket access allows users to see and list all objects in a bucket. This exposes object metadata details (for example, key and size) to users even if the users don't have permissions for downloading the object.

Amazon S3 Block Public Access must be disabled on the bucket

Confirm that there aren't any Amazon S3 Block Public Access settings applied to the bucket. These settings can override permissions that allow public read access. Amazon S3 Block Public Access settings can apply to individual buckets or AWS accounts.

If Requester Pays is enabled, then the request must include the request-payer parameter

If Requester Pays is enabled on a bucket, then anonymous access to the bucket is not allowed. Users from other accounts must specify the request-payer parameter when they send requests to the bucket. Otherwise, those users get an Access Denied error.

If you're using a Referer header to restrict access from CloudFront to your S3 origin, then review the custom header

If you're using the Referer header to restrict access from CloudFront to your S3 website endpoint origin, check the secret value or token set on the S3 bucket policy. Then, confirm that the secret value or token matches the value on the CloudFront origin custom header.

If you're using an explicit deny statement in the bucket policy, then confirm that there's also an allow statement that grants access based on the Referer header. You can't grant access with only an explicit deny statement.

For example, the following bucket policy grants access to the S3 origin when the request contains the string "aws:Referer":"MY_SECRET_TOKEN_CONFIGURED_ON_CLOUDFRONT_ORIGIN_CUSTOM_HEADER":

{
  "Version":"2012-10-17",
  "Id":"http referer policy example",
  "Statement":[
    {
      "Sid":"Allow get requests originating from my CloudFront with referer header",
      "Effect":"Allow",
      "Principal":"*",
      "Action":"s3:GetObject",
      "Resource":"arn:aws:s3:::DOC-EXAMPLE-BUCKET/*",
      "Condition":{
        "StringLike":{"aws:Referer":"MY_SECRET_TOKEN_CONFIGURED_ON_CLOUDFRONT_ORIGIN_CUSTOM_HEADER"}
      }
    }
  ]
}
With this example bucket policy, the CloudFront origin custom header must be:

  • Header: Referer
  • Value: MY_SECRET_TOKEN_CONFIGURED_ON_CLOUDFRONT_ORIGIN_CUSTOM_HEADER

Note: The example bucket policy grants public (anonymous) access to the bucket because the Principal is a wildcard value ("Principal":"*"). However, because of the condition statement, access to the S3 origin is granted only if the request includes the Referer header and the header value matches the value in the bucket policy.