On 8/30/2006 6:03 PM, ned+ietf-smtp(_at_)mrochek(_dot_)com wrote:
We eventually settled on an intermediate strategy. Temporary failures before
the MAIL FROM caiuse us to retry using the next MX, failures at or after the
MAIL FROM do not. We have found that this works pretty well overall.
Yeah. IMO it breaks down like this:
Failure to establish SMTP session, goto next host. Failure to negotiate
SMTP transfer (bad AUTH requirement, whatever), goto next host. Failure to
actually transfer the message (recipient unknown, SIZE problem, whatever),
respect the code and either requeue or abort accordingly.
An MX is essentially a promise to accept mail for a domain. If they are
lying about it or the domain is misconfigured, that's the recipient's
problem not the sender's. "Reliability" isn't about guaranteeing that
you'll overcome the recipient's inadequacies in life.
--
Eric A. Hall http://www.ehsco.com/
Internet Core Protocols http://www.oreilly.com/catalog/coreprot/