> I have a postfix(postfix 2.6.5) mail receiving server.
> On this I have used an email filter (sendmail-jilter http://sendmail-jilter.sourceforge.net/) that have some of my custom code. Following setting is used for the milter in 'main.cf':
> #Milter support for smtpd mail
> smtpd_milters =
> milter_protocol = 2
> Here on email loads of approx 40/hour the setup is fine but for email loads of approx 600/hour, I am facing following messages in postfix logs:
> Aug 9 07:34:12 ip-10-244-155-191 postfix/cleanup: warning: milter inet:localhost:10028: can't read SMFIC_EOH reply packet header: Success
> Aug 9 07:34:12 ip-10-244-155-191 postfix/cleanup: 771CB100168: milter-reject: END-OF-MESSAGE from xxx.atlanta.xx.com[15.xxx.32.xx]: 5.5.0 Service unavailable; from=<xxxx.xxxx@**.com> to=<[hidden email]> proto=ESMTP helo=<xxx.atlanta.xx.com>
Have you configured "milter_default_action = reject"? Don't do that
if you want mail to be tried again latrer.
Unfortunately, Milter applications can fail under high load - they
run inside one multi-threaded process, while Postfix runs one smtpd
process and one cleanup process per SMTP session.