feat(codedeploy): CodeDeploy deployment group construct for ECS

[clareliguori]

closes #1559

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license

[gitpod-io]

[clareliguori]

This PR should replace #22081 (thanks @cplee!)

[clareliguori]

@corymhall, I attempted to add an integ test that actually starts a deployment as we discussed in PR #22159, but ran into some issues. You can see the diff here showing the test:
clareliguori@70ed0ae

I ran into a few issues:

1. I had expected the test would retry the getDeployment call for a while until receiving the expected output (status = successful). It appears that the stack only calls getDeployment once and then checks the output in a different CFN resource? So the test gives up while the deployment is still in progress.
2. It appears that the API call custom resource calls the API on every event (create, update, and delete), so while the stack was deleting, it attempted to create yet another deployment by calling createDeployment on delete.
3. The output of getDeployment is sometimes too big, so I ran into (integ-tests): awsApiCall Response object is too long #22059

Thoughts? Is there another pattern I should follow here? I did confirm that the deployment started by the integ test does eventually succeed, but ideally this would run every time as part of the integ test suite.

[corymhall]

• I had expected the test would retry the getDeployment call for a while until receiving the expected output (status = successful). It appears that the stack only calls getDeployment once and then checks the output in a different CFN resource? So the test gives up while the deployment is still in progress.

Yeah that is something that I'm working on. Once #22238 I have another PR ready to add waiting for a certain result.

• It appears that the API call custom resource calls the API on every event (create, update, and delete), so while the stack was deleting, it attempted to create yet another deployment by calling createDeployment on delete.

This should get fixed in #22238.

• The output of getDeployment is sometimes too big, so I ran into

I don't have a fix yet for this, but I have a couple of ideas.

Thoughts? Is there another pattern I should follow here? I did confirm that the deployment started by the integ test does eventually succeed, but ideally this would run every time as part of the integ test suite.

I think for now as long as you have run the test and manually confirmed that is the best we can do. Just make sure to add a comment block to the integration test detailing the manual steps to follow. Once some of these integ test enhancements land we can revisit.

[corymhall]

I've done an initial pass. Is this a new feature that was recently released? I know people have been
asking for this for a while and I haven't seen any announcement or documentation that explains how
to do this.

My biggest concern right now is that while you can configure the actual deployment, you can't
actually perform the deployment? I know that is the first question that people will ask. Unless I am
missing something.

[cplee]

My biggest concern right now is that while you can configure the actual deployment, you can't
actually perform the deployment? I know that is the first question that people will ask. Unless I am
missing something.

@corymhall - I have a PR that will follow right after this one to add support for ECS deployments. Would it help to draft that PR now?

[clareliguori]

I've done an initial pass. Is this a new feature that was recently released? I know people have been asking for this for a while and I haven't seen any announcement or documentation that explains how to do this.

This feature was released in 2018, but the CloudFormation release notes show that CloudFormation support for the new attributes in AWS::CodeDeploy::DeploymentGroup was added July 8, 2021. I think no one noticed when CFN support was added? ¯\(ツ)/¯.

My biggest concern right now is that while you can configure the actual deployment, you can't
actually perform the deployment? I know that is the first question that people will ask. Unless I am
missing something.

@cplee noted that he has a follow-up PR for doing in-stack deployments. Though, based on the comments in the original issue, most folks are looking to stand up all the resources in CFN, and then deploy to the service in a pipeline outside of CFN. The current alternative is to use this feature, which is already supported by CDK:
https://aws.amazon.com/about-aws/whats-new/2020/05/aws-cloudformation-now-supports-blue-green-deployments-for-amazon-ecs/

Pull request has been modified.

[corymhall]

@corymhall - I have a PR that will follow right after this one to add support for ECS deployments. Would it help to draft that PR now?

Yes that would be great! I think that would help me see the full picture

Pull request has been modified.

[cplee]

@rix0rrr @corymhall - it looks like all feedback and requested changes have been addressed on this PR. Is there anything else needed to get this PR approved?

(FYI - PR #22455 builds on this PR by adding a construct for creating CodeDeploy deployments. It is ready for review immediately following this one)

[corymhall]

@clareliguori sorry for the delay on this review! Thanks for removing the editing of the underlying
service. I think the only thing we need to add now is:

1. Validation that the service is configured correctly. I think this would just be that the
   deploymentController and taskDefinition are configured correctly.
2. Configuration of the service (done in the BaseService construct). I think this would just be
   that when the user configures deploymentController: DeploymentControllerType.CodeDeploy that
   the revision is stripped from the task definition.

Pull request has been modified.

[clareliguori]

@clareliguori sorry for the delay on this review! Thanks for removing the editing of the underlying service. I think the only thing we need to add now is:

1. Validation that the service is configured correctly. I think this would just be that the
   `deploymentController` and `taskDefinition` are configured correctly.

2. Configuration of the service (done in the `BaseService` construct). I think this would just be
   that when the user configures `deploymentController: DeploymentControllerType.CodeDeploy` that
   the revision is stripped from the task definition.

@corymhall these are both added now

[clareliguori]

Ping @corymhall - hopefully we're good to merge now 😄

[corymhall]

@clareliguori thanks for the updates. Just one last conversation to resolve!

Pull request has been modified.

[clareliguori]

@corymhall ah somehow I missed that one, thanks. Should be resolved now

Pull request has been modified.

[clareliguori]

@corymhall Whoops I forgot to update the samples in the readme - can you approve again?

[mergify]

Thank you for contributing! Your pull request will be updated from main and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).

[mergify]

Thank you for contributing! Your pull request will be updated from main and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).

[aws-cdk-automation]

AWS CodeBuild CI Report

• CodeBuild project: AutoBuildv2Project1C6BFA3F-wQm2hXv2jqQv
• Commit ID: 0637ba4
• Result: SUCCEEDED
• Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

[mergify]

Thank you for contributing! Your pull request will be updated from main and then merged automatically (do not update manually, and be sure to allow changes to be pushed to your fork).