Hi Gilles, The MX failure I reported does not occur with the latest version of OpenSMTPd.
I installed OpenBSD 5.3 onto a second machine, leaving all the defaults including smtpd.conf, pf.conf. Mail was still stuck in the queue, with the Temporary MX lookup failure message. So we can rule out a misconfiguration of OpenBSD, but not the possibility of some strange misconfiguration of my own environment. I downloaded the latest version (opensmtpd-201310031056) make && make install, and started smtpd. No problem - mail flows through as you'd expect. So I'm ruling out a strange misconfiguration of my network environment. Installed the new version on the first machine, and everything also works fine on that machine too. I'm sharing this with the list in case anyone else experiences the temporary MX lookup issue in the future, although I'm afraid all I can suggest as a resolution is to upgrade. kind regards Richard P.S. love the recent syntax changes, especially pki! On 03/10/2013, at 5:40 AM, Richard Kernahan <[email protected]> wrote: > Hi Gilles, > Thank you for the suggestion. > I'm afraid I have some obscure misconfiguration, most likely in pf.conf > I intend to rebuild the machine from scratch and add the rules one by one > until I break it again :-) > If any insights relevant to OpenSMTPd arise I'll report back to the list. > kind regards > Richard > > On 30/09/2013, at 6:33 PM, Gilles Chehade <[email protected]> wrote: > >> On Tue, Sep 24, 2013 at 11:00:35PM +1000, Richard Kernahan wrote: >>> Hi All, >>> OpenSMTPd from the OpenBSD 5.3 i386 release is installed, and outgoing mail >>> is stuck in the queue, unable to find the MX server to relay to. What I >>> find confusing is that some emails have been relayed (after some delay). >>> >>> tcpdump reports that smtpd is sending 3 UDP messages to the name server, >>> but when the reply arrives for the first request, an ICMP port unreachable >>> message is returned, indicating the first port is unreachable. >>> I suspected pf (I always suspect pf :-) but when using dig, we see the MX >>> query and the UDP reply is accepted. The difference I see is that dig sends >>> only one request instead of 3. >>> In both cases (smtpd and dig) `systat states` shows the UDP states. >>> It *looks* like smtpd is no longer listening for a response on the first >>> request's port. I don't see a response to the second and third requests. >>> >>> Does anything in the following configuration appear wrong or questionable? >>> Does smtpd making 3 DNS requests seem reasonable? What might cause it to >>> stop listening for a response? >>> >> >> Sorry for the delay, I forgot about your mail :-) >> >> Does it still happen if you add domain to the tcpserv macro in your pf.conf ? >> >> -- >> Gilles Chehade >> >> https://www.poolp.org @poolpOrg > -- You received this mail because you are subscribed to [email protected] To unsubscribe, send a mail to: [email protected]
