Both clusters are set up with MDC replication.
In a nutshell, once a file is replicated from DC "A" to the remote DC "B" it is not possible to retrieve the file from the bucket with s3cmd (will result in an s3cmd error).
Pulling the file out with curl works but completely missing all HTTP headers and truncated by 3 bytes.
This is 100% reproducible on both of our MDC clusters.
Regards,
Peter