We thought of this! As francislavoie said, Caddy's retry logic involves switching to the Let's Encrypt staging endpoint until the DNS resolves properly. That is

We thought of this! As francislavoie said, Caddy's retry logic involves switching to the Let's Encrypt staging endpoint until the DNS resolves properly. That is coupled with exponential backoff. Hitting the staging endpoint as a test isn't a perfect thing, by the way -- it can help where DNS records are misconfigured, but for things like rate limits or other state internal to the CA server, switching endpoints will only give you a false positive. So, better to not hit that problem in the first place, and to make sure your DNS records are configured properly first. You can read more about these best practices here: https://docs.https.dev/acme-ops#avoid-using-a-production-ca-endpoint-while-testing-or-developing https://docs.https.dev/acme-ops#avoid-using-a-production-ca-...