Create custom domain names that your lambda can deploy to with serverless. Allows for base path mapping when deploying and deletion of domain names.

Amplify builds innovative and compelling digital educational products that empower teachers and students across the country. We have a long history as the leading innovator in K-12 education - and have been described as the best tech company in education and the best education company in tech. While others try to shrink the learning experience into the technology, we use technology to expand what is possible in real classrooms with real students and teachers.

Learn more at https://www.amplify.com

Make sure you have the following installed before starting:

• nodejs
• npm
• serverless

The IAM role that is deploying the lambda might need the following permissions:

acm:ListCertificates                   *
acm:DescribeCertificate                *
apigateway:AddCertificateToDomain      /domainnames*
apigateway:RemoveCertificateFromDomain /domainnames*
apigateway:GET                         /domainnames*, /apis*, /restapis*
apigateway:DELETE                      /domainnames*, /apis*, /restapis*
apigateway:POST                        /domainnames*, /apis*, /restapis*
apigateway:PATCH                       /domainnames*, /apis*, /restapis*
cloudformation:GET                     *
cloudformation:ListStacks              *
cloudformation:DescribeStacks          *
cloudfront:UpdateDistribution          *
route53:ListHostedZones                *
route53:ChangeResourceRecordSets       hostedzone/{HostedZoneId}
route53:GetHostedZone                  *
route53:ListResourceRecordSets         *
iam:CreateServiceLinkedRole            arn:aws:iam::${AWS::AccountId}: role/aws-service-role/ops.apigateway.amazonaws.com/AWSServiceRoleForAPIGateway
s3:ListBucket                          *
s3:GetObject                           *

Alternatively you can generate an least privileged IAM Managed Policy for deployment with this:

deployment policy cloudformation template

# From npm (recommended)
npm install serverless-domain-manager --save-dev

Then make the following edits to your serverless.yaml file:

Add the plugin.

plugins:
  - serverless-domain-manager

Add the plugin configuration (example for serverless.foo.com/api). For a single domain and API type the following structure can be used.

custom:
  customDomain:
    domainName: serverless.foo.com
    stage: ci
    basePath: api
    certificateName: '*.foo.com'
    createRoute53Record: true
    createRoute53IPv6Record: true
    endpointType: REGIONAL
    securityPolicy: tls_1_2
    apiType: rest
    autoDomain: false

Multiple API types mapped to different domains can also be supported with the following structure. The key is the API Gateway API type.

custom:
  customDomain:
    rest:
      domainName: rest.serverless.foo.com
      stage: ci
      basePath: api
      certificateName: '*.foo.com'
      createRoute53Record: true
      createRoute53IPv6Record: true
      endpointType: REGIONAL
      securityPolicy: tls_1_2
    http:
      domainName: http.serverless.foo.com
      stage: ci
      basePath: api
      certificateName: '*.foo.com'
      createRoute53Record: true
      createRoute53IPv6Record: true
      endpointType: REGIONAL
      securityPolicy: tls_1_2
    websocket:
      domainName: ws.serverless.foo.com
      stage: ci
      basePath: api
      certificateName: '*.foo.com'
      createRoute53Record: true
      createRoute53IPv6Record: true
      endpointType: REGIONAL
      securityPolicy: tls_1_2

Or for multiple domains

custom:
  customDomains:
    - http:
        domainName: http-api-${opt:RANDOM_STRING}.${env:TEST_DOMAIN}
        basePath: ''
        endpointType: REGIONAL
    - http:
        domainName: http-api-${opt:RANDOM_STRING}.${env:TEST_DOMAIN}.foo
        basePath: ''
        endpointType: REGIONAL

For multi-region deployments, a route53Params structure can be used to support latency or weighted routing policies

custom:
  customDomain:
    domainName: serverless.foo.com
    stage: ci
    basePath: api
    certificateName: '*.foo.com'
    createRoute53Record: true
    endpointType: REGIONAL
    securityPolicy: tls_1_2
    route53Params:
      routingPolicy: latency

To create the custom domain:

serverless create_domain

To deploy with the custom domain:

serverless deploy

To remove the created custom domain:

serverless delete_domain

Creating the custom domain takes advantage of Amazon's Certificate Manager to assign a certificate to the given domain name. Based on already created certificate names, the plugin will search for the certificate that resembles the custom domain's name the most and assign the ARN to that domain name. The plugin then creates the proper A Alias and AAAA Alias records for the domain through Route 53. Once the domain name is set it takes up to 40 minutes before it is initialized. After the certificate is initialized, sls deploy will create the base path mapping and assign the lambda to the custom domain name through CloudFront. All resources are created independent of CloudFormation. However, deploying will also output the domain name and distribution domain name to the CloudFormation stack outputs under the keys DomainName and DistributionDomainName, respectively.

In version 3, we decided to create/update/delete all resources through the API. Previously, only the basepath mapping was managed through CloudFormation. We moved away from creating anything through the stack for two reasons.

1. It seemed cleaner to have all resources be created in the same fashion, rather than just having one created elsewhere. Since multiple CloudFormation stacks can't create the same custom domain, we decided to have everything be done through the API.

2. We ran into issues such as #57 where the CloudFormation wasn't always being applied.

However, we still add the domain name and distribution domain name to the CloudFormation outputs, preserving the functionality requested in #43 implemented in #47.

To run unit tests:

npm test

To run integration tests, set an environment variable TEST_DOMAIN to the domain you will be testing for (i.e. example.com if creating a domain for api.example.com). And ROUTE53_PROFILE for creating route53 record in one AWS account and deploy in another. Then,

export TEST_DOMAIN=example.com
export ROUTE53_PROFILE=default
export TLS_TRUSTSTORE_URI=s3://bucket-name/key-name
export TLS_TRUSTSTORE_VERSION=default

npm run build
npm run integration-test

All tests should pass. All unit tests should pass before merging. Integration tests will take an extremely long time, as DNS records have to propogate for the resources created - therefore, integration tests will not be run on every commit.

If there is an error update the node_modules inside the serverless-domain-manager folder:

npm install

Unit tests are found in test/unit-tests. Integration tests are found in test/integration-tests. Each folder in tests/integration-tests contains the serverless-domain-manager configuration being tested. To create a new integration test, create a new folder for the handler.js and serverless.yml with the same naming convention and update deploy.test.ts or create a separate one with the test.ts ending.

AWS API Gateway has three different API types: REST, HTTP, and WebSocket. Special steps need to be taken when migrating from one api type to another. A common migration will be from a REST API to an HTTP API given the potential cost savings. Below are the steps required to change from REST to HTTP. A similar process can be applied for other API type migrations.

REST to HTTP

1. Confirm the Domain name is a Regional domain name. Edge domains are not supported by AWS for HTTP APIs. See this guide for migrating an edge-optimized custom domain name to regional.
2. Wait for all DNS changes to take effect/propagate and ensure all traffic is being routed to the regional domain name before proceeding.
3. Make sure you have setup new or modified existing routes to use httpApi event in your serverless.yml file.
4. Make the following changes to the customDomain properties in the serverless.yml confg:

   endpointType: REGIONAL
   apiType: http
   allowPathMatching: true # Only for one deploy

5. Run sls deploy
6. Remove the allowPathMatching option, it should only be used once when migrating a base path from one API type to another.

NOTE: Always test this process in a lower level staging or development environment before performing it in production.

• (5/23/2017) CloudFormation does not support changing the base path from empty to something or vice a versa. You must run sls remove to remove the base path mapping.
• (1/17/2018) The create_domain command provided by this plugin does not currently update an existing Custom Domain's configuration. Instead, it only supports updating the Route 53 record pointing to the Custom Domain. For example, one must delete and recreate a Custom Domain to migrate it from regional to edge or vice versa, or to modify the certificate.
• (8/22/2018) Creating a custom domain creates a CloudFront Distribution behind the scenes for fronting your API Gateway. This CloudFront Distribution is managed by AWS and cannot be viewed/managed by you. This is not a bug, but a quirk of how the Custom Domain feature works in API Gateway.
• (2/12/2019) Users who upgraded from 2.x.x to version 3.0.4 (now unpublished) and then reverted back to 2.x.x will be unable to deploy because of a bug that will be fixed in 3.1.0. The workaround is to delete the basepath mapping manually, which will let them successfully revert back to 2.x.x.
• (1/20/2022) Using route53Profile option requires having hosted zone for the domain in this profile and ACM certificate in the main profile (where functions are deployed).
• (2/13/2024) ACM certificate must exist in the us-east-1 for the EDGE endpoint type. https://docs.aws.amazon.com/apigateway/latest/developerguide/how-to-edge-optimized-custom-domain-name.html

If you have any security issue to report, contact project maintainers privately. You can reach us at github@amplify.com

We welcome pull requests! For your pull request to be accepted smoothly, we suggest that you:

1. For any sizable change, first open a GitHub issue to discuss your idea.
2. Create a pull request. Explain why you want to make the change and what it’s for. We’ll try to answer any PR’s promptly.