There was an issue with our database cluster causing this. This has been resolved now. We're still investigating the cause.