bottlerocket-test-system
bottlerocket-test-system copied to clipboard
Proposal: cluster creation failure should delete the CF stack
Currently, if cluster creation fails for some reason, the agent's eksctl
won't roll back the CF stack that was created. This can lead to latter results where subsequent cluster creations fail because the stack name is already taken:
2023-05-09 21:47:58 [✖] creating CloudFormation stack "eksctl-x86-64-aws-k8s-124-cluster":
operation error CloudFormation: CreateStack, https response error StatusCode:
400, RequestID: 3aedd327-554f-418c-8b7b-581c217787c0, Error: failed to create cluster "x86-64-aws-k8s-124"
Rollback completes succesfully but that stack name is still there. Not too hard to go in and manually clean up that stack, but would be great if this was alittle more 1-button-push