This was a problem on our upstream provider’s network and has now been resolved.
This issue was opened retrospectively.