Issues when terminating a stack

[vkurup]

I noticed the following issues when terminating an EB stack.

• 3 resources are not deleted and the event logs say that deletion was skipped: ContainerLogs (CloudWatch), PrivateAssetsBucket (S3), AssetsBucket (S3). The ContainerLogs have an expiration of 365 days, so they will eventually get deleted, and maybe it's a useful thing to have logs around for a while after termination. But the S3 buckets seem like they would just linger forever causing unneeded charges.
• 1 resource deletion fails, which causes the stack's status to be 'DELETE_FAILED'. This is the ECR which fails with this messsage: "The repository with name 'disco-fred' in registry with id '112250541543' cannot be deleted because it still contains images"

[vkurup]

Looks like the S3 buckets and Logs are intentionally NOT deleted.

Some background: https://aws.amazon.com/blogs/devops/delete-your-stacks-but-keep-your-data/

This also led me to identify that the default policy for RDS is to save a snapshot before deletion. (Unfortunately these snapshots don't get tagged with our custom tags, so I didn't see them until reading about the DeletionPolicy docs)

So my suggestions:

1. Add docs about what resources are expected to remain after stack termination, and how to find them.
2. (Maybe) Add a user parameter that allows you to specify that everything should be deleted on termination.
3. Break out the ECR deletion fail as a separate bug (which I haven't been able to diagnose yet)

Thoughts?