Viktor - you nailed it. Not certain why just yet, but for a certain subset of
hosts, iptables is failing to pass packets containing the mail from:
line. After a certain number of (TCP) retries, some will be successful.
nf_conntrack isn’t hitting its limits or anything obvious (far from it), s
There is a regular ol’ iptables/linux box in front of this doing nat and state
matching, didn’t think to poke around there. Thanks for the tip.
Brian
On Apr 16, 2014, at 2:47 PM, Viktor Dukhovni wrote:
>>
>
> Some firewall or load-balancer or similar device is losing connection
> state on th
ostfix/smtpd[12387]: private/anvil: wanted attribute:
(list terminator)
Apr 16 13:37:20 mail postfix/smtpd[12387]: input attribute name: (end)
Apr 16 13:37:20 mail postfix/smtpd[12387]: timeout after EHLO from
vms173009pub.verizon.net[206.46.173.9]
Apr 16 13:37:20 mail postfix/smtpd[12387]: disconnect f
On Wed, Apr 16, 2014 at 02:17:23PM -0500, Brian Grimal wrote:
> 14:04:31.425682 IP 172.16.5.8.25 > 206.46.173.11.19129: Flags [P.], seq
> 35:216, ack 32, win 46, options [nop,nop,TS val 386720906 ecr 1770681376],
> length 181
> ..&.&mO-..J.
> i.t 250-mail.A.com
> 250-PIPELINI
Brian Grimal:
> Apr 16 12:57:45 mail postfix/smtpd[18647]: connect from
> vms173001pub.verizon.net[206.46.173.1]
> Apr 16 13:02:45 mail postfix/smtpd[18647]: timeout after EHLO from
> vms173001pub.verizon.net[206.46.173.1]
Postfix works as expected. Postfix receives *NOTHING* f
vms173001pub.verizon.net[206.46.173.1]
Apr 16 13:02:45 mail postfix/smtpd[18647]: timeout after EHLO from
vms173001pub.verizon.net[206.46.173.1]
Apr 16 13:02:45 mail postfix/smtpd[18647]: disconnect from
vms173001pub.verizon.net[206.46.173.1]
gmail, however, delivers quite reasonably:
--
Apr 16 13:54:29
Brian Grimal:
> after ehlo. I?ve telnetted into the box from off-site (regular
> host, not a mail server), and it immediately connects, immediately
> responds with the 220 mail..com ESMTP Postfix herald, and
> immediately responds to the helo or ehlo. However, it will take
> about two minutes
I’m banging my head on the desk over this one. Some hosts, fly right on
through - connect, ehlo, mail from, rcpt-to, data, quit .. done. All in maybe
1 second. Others, will consistently timeout after ehlo. I’ve telnetted into
the box from off-site (regular host, not a mail server), and it
