Mistakes will happen, sure. But given the following: > The replication procedure is super fragile, prone to error, relies on a handful of random shell scripts,

Mistakes will happen, sure. But given the following: > The replication procedure is super fragile, prone to error, relies on a handful of random shell scripts, and is badly documented > Our backups to S3 apparently don’t work either: the bucket is empty It seems like a lot of their backup and restore procedures were never even tested.