Just a quick reply to keep this thread open. We have a cloud server and have been having this issue as well and, as far as we can tell, the issue is not resolved. Certificates continue to fail on an almost daily basis. We find that the certificates that fail the most are those that use parked domains and those that use subdomains. For parked domains it’s not clear why they fail. For subdomains we believe it is because the SG solution requires both a bare domain and a www version, even if it is for a subdomain (which we don’t understand as www is a subdomain).
Indeed, manual intervention is the only solution at present. It is annoying and causes sometimes extended downtime.