search

aws / containers-roadmap

This is the public roadmap for AWS container services (ECS, ECR, Fargate, and EKS).
https://aws.amazon.com/about-aws/whats-new/containers/
Other
5.22k stars 320 forks source link

ECS: Blue/Green Service Deploy doesn't add Service's CapacityProviderStrategy to AppSpec #1752

Open eyedean opened 2 years ago

eyedean commented 2 years ago

I am coming from https://github.com/aws/containers-roadmap/issues/713#issuecomment-1133745505. It's a year old bug that I found a workaround for it (creating AppSpec manually and kicking the CodeDeploy deploy manually) but why is the bug there doesn't get fixed?

Since that issue is closed, I am creating a new one.

image
jensenbox commented 2 years ago

I stand on the shoulders of giants on this one. Had I not had the work of the previous brave souls I would never have found the solution to this bug.

I will say, this is a pretty serious issue with respect to the idea that you should be able to deploy code using something like Terraform and have it be stateful. Whenever I went to deploy a second time I would find that the Capacity Provider was removed and I had drift in my Terraform.

I would hope that we could get this bug squashed relatively soon as it does cause issues when using things like Terraform.

Essentially, CodeDeploy is removing the CapacityProviderStrategy from the service whenever it deploys and your AppSpec did not declare a CapacityProviderStrategy - in my mind, it should read whatever is there and use it.

I would definately classify this as a bug because it does use the Capacity Provider the first time it is deployed but deletes it after the run.

EmViKay commented 2 years ago

the same issue, please fix ASAP

yahya-uddin commented 1 year ago

How is this issue still open after all these years? Very big bug.

rotem925 commented 1 week ago

I was wondering the same. This need to be fixed as we cannot use ECS with blue-green and auto-scaling

pallymore commented 5 days ago

Hi all - this appears to have fallen through the cracks

We are actively working on fixing it in the new ECS Console now and will share an update as soon as the fix is shipped.

We apologize for not catching this sooner. We're thankful for your diligence in reporting it.