Bug: Use existing S3 bucket if stack is redeployed in prod

[Jimlinz]

Geostore provisions several s3 buckets (prod and staging) within its stack. These buckets are town down in non-prod environments when the cdk destroy command is issued, but they are set to retain in production. This is by design and also an expected behaviour.

However, should we need to destroy the stack in production one day and redeploy it, we would run into problems during (re)deployment since the bucket is already in place. We would likely get a resource already exist error when we try to (re)deploy the stack in production.

We should investigate the canonical way of dealing with such situation and update our stack accordingly.

Things to keep in mind:

• Bucket permission (i.e. how it could be made available to the new stack)

[Jimlinz]

Would this also apply to resources like dynamodb?

[Jimlinz]

Is it a good idea to do a periodic tear down the production stack and redeploy (perhaps in a different aws account) to capture potential problems?