Rob Arends wrote: > Yes I agree, if there is a lookup failure it should fail immediately. > Also if the DNS lookup is ok, and Xmail cannot connect to the IP address, it > should fail immediately. > > Now I hear you say, no!!! > > This is the correct function. This is why the RFCs call for a secondary > email server on a separate class C. > Although the separate class C bit is out of date due to CIDR and subnetting. > > But when the primary MX is uncontactable, an MTA should try the other MXs in > order until exhausted. > If no response from any of them, then 5xx error, immediate NDR.
No. Consider: Your mail server (locally run, for a few small domains) relays through your ISP's mail server for various reasons. Two of the major backbone providers have another of their periodic "wars", where they each refuse to carry some or all of the other's traffic (before you say it can't happen, it already has, several times - UUNet is shunned in several places, as is C&W). This has the effect of temporarily (these "wars" seldom last more than a day, usually only a few hours) isolating you from portions of the internet - those hosts are there, but the packets are not transiting between the two backbone providers needed to get your packets there. Now, if you implemented your "DNS lookup succeeds (in this case, perhaps, using cached data), but connection to MX fails, so perm-fail the message", you will be permanently failing messages to potentially a large number of hosts, thus causing problems for your users. And if everyone did this, people from both sides of the "war" would be having customers complaining. No... Much better to follow the old maxim "be liberal in what you accept, and conservative in what you send" - allow for failures that might not be within the letter of the RFC but could represent "live events" that were not anticipated or covered by the RFC. After all, it's not like a delay of a few hours (or even a day or so) is going to matter in the case of an NDN. Email is not, and was never intended to be, a time critical method of communication - if something is time sensitive, it needs to be communicated using another method. Now, with all that said, I do agree that immediate failure in the case of "NXDOMAIN" lookups would be appropriate. Although I can conceive of situations in which that would also represent a "temporary" failure condition. - To unsubscribe from this list: send the line "unsubscribe xmail" in the body of a message to [EMAIL PROTECTED] For general help: send the line "help" in the body of a message to [EMAIL PROTECTED]
