Prevent CloudFormation Change Sets from piling up

Andreas Wittig – 10 Oct 2019

Recently, I’ve stumbled upon a problem when using aws cloudformation deploy within deployment pipelines (Jenkins, GitLab CI, …) that I wanted to share with you.

Prevent CloudFormation Change Sets from piling up

Usually, I’m using the AWS CLI to deploy CloudFormation stacks.

aws cloudformation package --template-file example.yml --output-template-file output.yml --s3-bucket example
aws cloudformation deploy --template-file output.yml --stack-name example --no-fail-on-empty-changeset

The aws cloudformation package command packages the template example.yml and uploads dependencies like nested stack templates to S3.

The aws cloudformation deploy command creates or updates the stack example. The option --no-fail-on-empty-changeset makes sure the command does not throw an error in case the template has not changed, which is very likely when using a deployment pipeline.

It is essential to know that the pipeline creates a change set whenever executing aws cloudformation deploy. Deploying a stack without making any changes to the template or the parameters leads to a failed change set: The submitted information didn't contain changes. Submit different information to create a change set.

Andreas and Michael Wittig

Please support our work!

We have published 327 articles, 41 podcast episodes, and 15 videos. It's all free and means a lot of work in our spare time.

Thanks to Alan Leech, Alex DeBrie, e9e4e5f0faef, Goran Opacic, jhoadley, Shawn Tolidano, Thorsten Hoeger, Todd Valentine, Vince Fulco, and all anonymous supporters for your help! We also want to thank all supporters who purchased a cloudonaut t-shirt. It gives us great pleasure to send our t-shirts all over the world.

With your help, we can continue to produce independent & high-quality content focused on AWS. Please support us!

Support us

Unfortunately, the aws cloudformation deploy command does not clean up those failed change sets. This leads to change sets piling up. And one day or another your deployment pipeline will fail with the following error message:

An error occurred (LimitExceededException) when calling the CreateChangeSet operation: ChangeSet limit exceeded for stack ...

You have reached the maximum number of change sets for a stack.

How to fix that? It would be great if aws cloudformation deploy would clean up failed change sets automatically. Keep an eye on #4534 for progress on that.

For now, adding the following bash script to your deployment pipeline might help. The cleanup() function list all stacks starting with prefix, fetches the change sets for each stack, and deletes change sets in status FAILED.

# cleanup (region, prefix)
cleanup () {
stacks=$(aws cloudformation list-stacks --stack-status-filter CREATE_COMPLETE UPDATE_COMPLETE --query "StackSummaries[?starts_with(StackName, \`$2\`) == \`true\`].StackName" --output text --region $1)
for stack in $stacks
echo "${stack}: cleaning up change sets"
changesets=$(aws cloudformation list-change-sets --stack-name $stack --query 'Summaries[?Status==`FAILED`].ChangeSetId' --output text --region $1)
for changeset in $changesets
echo "${stack}: deleting change set ${changeset}"
aws cloudformation delete-change-set --change-set-name ${changeset} --region $1

The following command executes the function cleanup() for region eu-west-1 and stack name prefix example-.

cleanup eu-west-1 example-

In summary, aws cloudformation deploy leaves behind waste that you need to deal with.

Andreas Wittig

Andreas Wittig

I launched in 2015 with my brother Michael. Since then, we have published hundreds of articles, podcast episodes, and videos. It’s all free and means a lot of work in our spare time. We enjoy sharing our AWS knowledge with you.
Have you learned something new by reading, listening, or watching our content? If so, we kindly ask you to support us in producing high-quality & independent AWS content. We look forward to sharing our AWS knowledge with you.

Support us

Feedback? Questions? You can reach me via Email, Twitter, or LinkedIn.