AWS::AutoScaling::AutoScalingGroup - UpdatePolicy does not work with AutoScalingGroup Mixed Instance Policy

RyanFrench commented 4 years ago

1. AWS::AutoScaling::Group

2. Scope of request

AWS:: AutoScaling:: Group::MixedInstancePolicy causes update failures with UpdatePolicy when a server is launched with Weighted Capacity > 1, and using the Resource Signals feature.

3. Expected behavior

When updating a stack with a Mixed Instance Policy, if any instance is launched with a weighted capacity > 1, CloudFormation should take this into account when waiting for resource signals.

Feel free to describe other desired and relevant implementation details.

4. Suggest specific test cases

1.

Deploy a stack with 12 m5.large instances using CloudFormation, and a weighted capacity of 1, with an UpdatePolicy that specifies waiting for resource signals.
Deploy an update to this stack, switching the 12x m5.large to 6x m5.xlarge, with a weighted capacity of 2.
The stack will fail to update as it expects 12 resource signals to be sent, even though only 6 new servers have been started by the ASG.

5. Helpful Links to speed up research and evaluation

6. Category (required) - Will help with tagging and be easier to find by other users to +1

Compute (EC2, ECS, EKS, Lambda...)

7. Any additional context (optional)

mrapczynski commented 4 years ago

@luiseduardocolon We have experienced something very similar to this when using the mixed instances policy on an autoscaling group. I opened a support case on in 2019, and the AWS representatives confirmed the bug, but did not provide a timeline for the fix. See case # 5814859491 if you have access to the histories.

The workaround for now has been change the resource name of the ASG in the template, forcing CF to create a new autoscaling group with the relevant configuration updates, and the old one is discarded.

BugDestroy3r commented 4 years ago

@mrapczynski have you got any update on this bug-fixes?

trying to debug for hours until seeing this and the workaround is not realistic for the stack that frequently changed and/or under active development.

mrapczynski commented 4 years ago

@BugDestroy3r No, as far I know, this is still a problem. I do not work for AWS, and have been experiencing this as a customer.

taraspos commented 3 years ago

So, I've reported this to AWS Support and they told me that they were able to reproduce the issue, but "could not provide an ETA on when the fix will be deployed" 🤷‍♂️

aws-cloudformation / cloudformation-coverage-roadmap