Victor Duchovni wrote: > On Thu, Apr 16, 2009 at 06:47:58PM -0700, Seth Mattinen wrote: > >> I apologize in advance if I'm being horribly dense, but I'm seeing >> something that doesn't feel right. In the event that a transport map >> lookup fails with a "host not found" error, Postfix is bouncing the >> message rather than treating it as a temporary error. > > The transport map lookup did not fail. > >> For my test, I have the transport map: >> >> 50lightyears.com smtp:[badrecord.mattinen.org]:1234 >> >> Where "badrecord.mattinen.org" intentionally does not exist. So I send a >> test message and I see this in the logs: >> >> postfix/smtp[5361]: 773FA3E442: to=<t...@50lightyears.com>, relay=none, >> delay=0.25, delays=0.21/0.04/0/0, dsn=5.4.4, status=bounced (Host or >> domain name not found. Name service error for >> name=badrecord.mattinen.org type=A: Host not found) > > The destination nexthop does not exist, this is not a transient condition.
In my example yes (I just needed a way to force the behavior, that's not a real application), but in the issue I'm attempting to resolve the error was: "Host or domain name not found. Name service error for name=mail.xxxxx.net type=A: Host found but no data record of requested type" I assume - but haven't confirmed yet - this was caused by a Dynamic DNS hostname in the transport. I assume the provider withdrew the A record until the updated IP was available but during this time Postfix bounced anything heading for that transport. >> So, my question is, why is that fatal instead of temporary? Shouldn't it >> be temporary? Observed on 2.5.5 and 2.4.5. > > It should not be temporary. All lookups succeed and establish that the > destination is non-existent. Postfix correctly bounces the message. > > If you really want sub-optimal behaviour and a queue full of junk, try: > > smtp_defer_if_no_mx_address_found = yes > So am I correct in assuming that any lookup failure (aside from DNS timed out) at the transport map stage will result in a fatal condition? ~Seth
