Re: [atlas] Exact semantics of DNS' use_probe_resolver