Wieste, I appreciate your offer to take a look at a pcap but must decline due to the sensitivity of the contents. I did take a look at some pcaps and noticed tcp rst's were almost randomly being sent from different milters to postfix causing the incomplete conversation leading to the errors. I didn't spend much time digging around trying to find the reason for the tcp rcts before I jumped ship and changed my mail cluster to using unix sockets. Since the change i've not observed any SMFIC errors in the maillog. This leads me to believe that the problem is some where in the OS, possible something that could be ironed out with tuning. I wish I had more to offer in terms of a cause / solution, but for now my problems have been solved.
Thank you all for your time and advice =) -Nick On Fri, Jun 19, 2015 at 11:48 AM, Wietse Venema <wie...@porcupine.org> wrote: > Nick Winn: >> I've removed all of the extras possible from my main.cf and am running >> the most basic postfix config now. >> >> Here is the output of postconf -n >> >> https://paste.fedoraproject.org/234332/26834143/ >> >> I am still seeing the errors in the log files. =( >> >> What should my next troubleshooting steps be? > > As Viktor wrote, full packet recordings. > > # tcpdump -s 0 -w /file/name -i interface port xxx > > Where xxx is the port of the failing milter. I need to examine the > resulting file itself, not a printable "user friendly" version of > that file. Use direct email instead of the Postfix mailing list. > > My money is on a buggy Milter program. > > Wietse -- --- Nick Winn
