We have the remote_backup_user variable in some playbooks, it can be made more global for the "Create backup user" part.
All servers have been provisioned with the changes. Let's wait another night to see if the old machine still receives logs or backups from the other machines.
If we don't want to wait until Monday, the logrotate config in templates/logrotate/nginx-fundraising.j2 needs to be set from weekly to daily. This is probably a good idea anyway