Email monitoring introduced in #68 doesn't have any mechanism to cache/filter out repeated errors. With a few requests per minute now it's not a big issue, but can be an issue later if we accidentally break something in the production.