Why (the Hell) is Snapshot the CloudFormation RDS Default Deletion Policy??
This one blew me away. If you create an RDS instance using any other method, the default deletion behaviour is just to delete. If you create the same configured instance using CloudFormation, the default deletion behaviour is to create a manual snapshot.
I was happily working through an A Cloud Guru CloudFormation course, spinning up CFN stack after CFN stack. In between I was deleting the stacks, thinking this was cleaning up my environment. Little did I know, there were RDS snapshots being created each time I killed my “immutable” infrastructure.
So, after plugging away at the course, I found that the default Deletion Policy for resources created using CloudFormation is to delete… Except for RDS… Grrrr!
In order to ensure my test infrastructure was properly cleaned up after being done, I had to ensure the DeletionPolicy field was manually set to “Delete” for the RDS instances being created:
Resources:
DB:
Type: \"AWS::RDS::DBInstance\"
DeletionPolicy: Delete
Properties:
...The AWS documentation for Deletion Policies says that:
“The default policy is Snapshot for AWS::RDS::DBCluster resources and for AWS::RDS::DBInstance resources that don't specify the DBClusterIdentifier property.”
What’s crazy is that this isn’t evident unless you see the snapshots being created. The CFN events for the stack don’t show the snapshot being created, just the DB being deleted:
I can (kind of) understand why AWS would want a snap to be taken before deleting a stack. Data loss is likely a bigger issue for them than an extra storage cost would be. I just think they’d have done a better job making this default behaviour salient, especially for those just getting started with CloudFormation.
Alright, off to delete some RDS snapshots…