I was wondering if anyone has ever done backups to an s3 bucket that exists in a different AWS account? I’m running into issues with S3 Object ACLs. The files that are being uploaded from AWS Account A to the S3 Bucket in AWS Account B are restricted to the AWS account that uploaded the objects (in this case AWS Account A). The Bucket owner (AWS Account B) does not have permissions to access the files. I spoke with AWS Support and the --acl bucket-owner-full-control parameter needs to be included to give the bucket owner in AWS account B full access to those objects. I there any way to configure this when using the BACKUP command in memsql?
For your first question, currently we do not have the --acl bucket-owner-full-control parameter set when doing the backup. A workaround for this would be to have your policy set, so that any uploads to a bucket give full control to the bucket owner. Amazon has a good example of how to do this here: Bucket policy examples - Amazon Simple Storage Service
For the second, you can pass empty credentials, however syntactically you still need to provide a credentials parameter, you can do this by just passing ‘{}’ as the credential string.
Modifying the example found in the docs, it would look like this:
BACKUP DATABASE memsql_demo TO S3 "backup_bucket/backups/6_1_2018" CONFIG '{"region":"us-east-1"}' CREDENTIALS '{}'