Append account number to S3 bucket names #6860
Description
User Story
As a user deploying Sleeper, I want Sleeper's AWS resource names to not conflict with other accounts, so that my choice of instance ID will not conflict with other instances in other accounts.
Description / Background
S3 bucket names are globally unique, and a maximum of 63 characters long. This is one of the main restrictions on the length of the Sleeper instance ID, which we've set to a maximum of 20 characters.
We'd like to append the account number to all S3 bucket names deployed in a Sleeper instance, to avoid conflicts with similarly named Sleeper instances.
Technical Notes / Implementation Details
This will be for every S3 bucket deployed in the cdk and system-test-cdk modules.
Taking a long bucket name as an example:
sleeper-<instance-id>-bulk-export-results
If you append the account name and region you get something like this:
sleeper-<instance-id>-bulk-export-results-123456789012-eu-west-2
With a 20 character instance ID that comes to 72 characters.
More generally, buckets would be named in this form:
sleeper-<instance-id>-<name>-<account>-<region>
The "sleeper" and hyphens come to 11 characters.
The account name is exactly 12 digits long.
The longest region names are 14 characters long.
Overall, that leaves 63 - 20 - 11 - 12 - 14 = 6 characters for a name specific to the bucket.
We'll need to make sure that name portion is 6 characters or less for every bucket we deploy.
We can leave the region out of the name and just include the account number, which would take off -<region>, giving us 6 + 14 + 1 = 21 characters for the name portion. That seems worth doing, as it's unlikely you'll want to use the same instance ID in multiple regions.