>-----Message d'origine----- >De : [EMAIL PROTECTED] >[mailto:[EMAIL PROTECTED] la part de David Lord >Envoy=E9 : dimanche 27 avril 2008 17:52 >=C0 : [email protected] >Objet : [xmail] Re: multiple smtprelay servers for custdomains and 5xx >er rors > > >On 26 Apr 2008 at 22:46, Davide Libenzi wrote: > >> On Thu, 24 Apr 2008, Oliver St=F6neberg wrote: >> > > On Tue, 22 Apr 2008, CLEMENT Francis wrote: >> > > >> > > > Seems there was some post about this. >> > > > >> > > > The BIG figure is when the final domain have multiple >mx servers. >> > > > Suppose the final domain have two mx and one is >misconfigured and return >> > > > a 5xx. >> > > > So if xmail tries first the 'bad' server, what to do next ? >> > > > >> > > > On a atomic try/retry cycle, you have two choices : >> > > > 1 - don't try others mx and send back an NDR to the sender >> > > > 2 - Try the others reminding MX for this try/retry >cycle, and send back >> > > > an NDR ONLY AND ONLY IF all mx return a 5xx, else >schedule a normal >> > > > retry cycle (that will retry on each mx). >> > > > >> > > > I think option 2 is better :) >> > > > >> > > > Could Davide tell us how xmail handles the 5xx with >multiple mx ? >> > > >> > > As far as smtprelay goes, a failure (of whatever type), >means try the next >> > > server in the list. >> > >> > That's why I wrote a mail, because it makes no sense to >continue in >> > the case of a permanent error. >> > >> > Is there a way to configure XMail with multiple domains >and stop when >> > you get a permanent error? >> >> I remember there was a discussion or something, time ago. >But now I can't >> remember. It seems sane to me that a 5xx error stops the >processing, and >> I'll schedule this for change. >> Speak now or forever hold your peace, in case you rely on >XMail re-trying >> even upon 5xx ... > >Looking at RFC2821 I can't really make out what is intended to happen >in this situation. It seems ok to stop and send an NDR, or that the >server that gave the 5xx response should not be retried but other >servers MAY be tried. So I'd say go for option (2) as most robust >method. > > >David >
I agree as after re-read rfc2821 I find this : >>>> 4.2.5 Reply Codes After DATA and the Subsequent <CRLF>.<CRLF> .... snip to last paragraph When an SMTP server returns a permanent error status (5yz) code after the DATA command is completely with <CRLF>.<CRLF>, it MUST NOT make any subsequent attempt to deliver the message. As with temporary error status codes, the SMTP client retains responsibility for the message, but SHOULD not again attempt delivery to the same server without user review and intervention of the message. <<<< It seems that the smtp client (xmail here when sending to smtp gateways = or mx servers), when receiving a 5xx code SHOULD NOT again attempt to the same server. So Davide is right, xmail can try the other gateways/mx available in the list. But with the possible case (Oliver case here) that "all the = gateways/mxs return a 5xx error" the smtp client have no more options to send the mail (because it "SHOULD not again attempt delivery = to the same server" and allready tried all). So the smtp client (xmail here) have to send an ndr to ask/alert the = sender what to do next (to comply with "without user review and intervention = of the message") :) So option 2 I described later seems to be the good interpretation : On a atomic try/retry cycle : Try the others reminding MX (or gateways) for this try/retry cycle, = then send back an NDR ONLY AND ONLY IF all mx (or gateways) return a 5xx, = else schedule a normal retry cycle (that will retry on each mx). Francis - 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]
