I have a CDK Pipeline that creates a Docker Image using DockerImageAsset and then deploys it to Elasticbeanstalk via ECR using a Dockerrun.aws.json.
To make this work, I had to split the deployment into two, so that the EB portion can have the generated image URL from ECR passed in as context.
This value is not available during synth, so it requires a deploy followed by a synth of the next stack.
I created a shell script that runs the first deploy, captures the output, then runs the second deploy. This shell script is small but would need to grow to support additional parameters over time.
I'd like to go back to a solution where all the logic is contained within CDK stacks, to the point where a simple "cdk deploy --all" works correctly again.
What are my options?
