ECS express gateway deployment

Question

I have developed a backend service using Express Gateway and set up a continuous deployment pipeline using AWS CodePipeline integrated with Amazon ECS.

In the pipeline, I am using the ECS standard deployment approach by adding the native ECS deploy stage. I have configured the necessary permissions, including IAM PassRole and required ECS access.

As part of the build stage, I updated the buildspec.yml file to generate an imagedefinitions.json file. This file includes the image URI and container name required for the ECS deploy stage. Since Express Gateway does not provide an option to explicitly define the container name, I used the container name that was automatically generated for the service so that it can be referenced during deployment.

However, the pipeline is failing during the deploy stage with an “internal server error.”

I would like to understand whether this issue could be related to Express Gateway deployment, and if so, what alternative approaches can be considered for deploying the service reliably within this setup.

re:Post Agent · Answer

Based on the information available, there appears to be a fundamental misunderstanding about what Amazon ECS Express Mode services are and how they should be deployed.

Amazon ECS Express Mode (also called Express Gateway Service) is a simplified deployment model for containerized web applications on ECS that automatically manages AWS infrastructure components like Application Load Balancers, target groups, security groups, and auto-scaling policies. It's not related to the Express.js framework or Express Gateway API gateway software.

The deployment approach you're describing—using CodePipeline with a standard ECS deploy stage and imagedefinitions.json—is the traditional method for deploying ECS services. However, Express Mode services use a different architecture called "service revisions" rather than task definitions, which means the standard ECS deployment process with imagedefinitions.json may not be compatible.

For Express Mode services, you need to specify a primary container configuration that includes at minimum a container image. The service manages the infrastructure automatically, so you don't manually define container names in the same way as traditional ECS services.

If you're experiencing deployment failures with "internal server error," this could be due to several factors:

1. **IAM Permission Issues**: Verify that both the task execution role and infrastructure role have the necessary permissions. The infrastructure role is specific to Express Mode and manages AWS resources on behalf of the service.

2. **Incompatible Deployment Method**: The standard ECS deploy stage in CodePipeline may not support Express Mode services, which would explain the internal server error.

For reliable deployment of Express Mode services, you should use the CreateExpressGatewayService API operation or equivalent AWS CLI/SDK commands rather than the traditional ECS deployment pipeline. If you need to use a CI/CD pipeline, you may need to create a custom deployment stage that calls the appropriate Express Mode APIs instead of using the native ECS deploy stage.

If you're actually trying to deploy an Express.js or Express Gateway application (not using ECS Express Mode), then you should use the standard ECS service deployment approach without Express Mode, which would be compatible with your current CodePipeline setup.  
**Sources**  
[ECSExpressGatewayService - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/APIReference/API_ECSExpressGatewayService.html)  
[CreateExpressGatewayService - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/APIReference/API_CreateExpressGatewayService.html)  
[Troubleshooting Amazon ECS Express Mode services - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/developerguide/express-service-troubleshooting.html)

ECS express gateway deployment

Add your answer

Relevant content