On Wed, Apr 02, 2014 at 09:48:05AM +0530, KK Patnaik wrote:
> 1) [root@smtp2]#qshape incoming active deferred
> T 5 10 20 40 80 160
> TOTAL 95414 8130 5019 11477 13835 33457 23496
> conrepmail.com 16397 1871 606 1255 3142 6792 2731
> aol.com 743 30 45 83 97 308 180
> vsoftconsulting.com 356 10 12 49 43 146 96
> nihaki.com 308 8 9 24 47 130 90
> baanyan.com 277 7 8 26 42 115 79
> sbcglobal.net 270 11 6 30 37 109 77
> net.com 265 30 19 28 38 83 67
> erpanderp.com 253 4 11 29 32 109 68
> techgene.com 240 10 12 19 42 91 66
> stansource.com 230 7 11 35 29 85 63
> a1k.com 225 7 8 26 33 93 58
> canvasinfotech.com 222 8 7 33 32 84 58
> comsys.com 221 25 14 35 29 69 49
> rsrit.com 214 8 11 28 35 88 44
> usmsystems.com 213 4 13 26 24 87 59
> compunnel.com 211 8 9 25 28 91 50
Out of ~96000 still queued messages that arrived in a 2--3 hour
window, 1 in 6 is a bounce back to your own domain. If you're not
a spammer, your list management practices are extremely poor. If
your intent is to process mail for legitimate non-spamming customers,
you are not doing a good job of keeping the folks with garbage lists
away.
Furthermore, why are bounces back to your own domain not delivered
promptly? Your bounce rate should probably be well below 2%, not
16%. You should be using the "relay" transport to reach
conrepmail.com so that bounces don't queue behind outbound mail.
I asked for collated samples for delayed mail from some of the
above most delayed domains, below you show mail that was not
substantially delayed. You're wasting everyone's time presenting
tiny crumbs of information, and often the wrong crumbs.
There are still no "c+d" statistics (moving expontially decaying
average over time is best).
# If handling multi-recipient mail, process only the first of
# multiple log entries with same smtpd[<pid>] and same queue-id
# logged during the same minute.
#
foreach SMTP delivery # insert real code here
$meancd = 0.99 * $meancd + 0.01 * ($c + $d);
if ($count % 100 == 0) {
printf "%s %s\n", $delivery_time, $meancd;
}
> Mar 31 09:11:15 smtp2 postfix/smtpd[16466]: connect from
> mm3.conrep.com[76.12.143.99]
> Mar 31 09:11:15 smtp2 postfix/qmgr[10231]: 8872C12400DC:
> from=<uemlmm3_87001237348238_100130003...@conrepmail.com>, size=53529,
> nrcpt=1 (queue active)
No delay entering the active queue, and this is not one of the domains
at the top of the qshape listing AND:
> Mar 31 09:11:18 smtp2 postfix/smtp[15252]: 8872C12400DC:
> to=<ab...@promorphics.com>,
> relay=promorphics-com.mail.protection.outlook.com[207.46.163.138]:25,
> delay=2.9, delays=0.07/0/0.16/2.6, dsn=2.6.0, status=sent
> (250 2.6.0 <20140331131115.8872c1240...@smtp2.conrepmail.com>
> [InternalId=22514218568637,
> Hostname=DM2PR04MB527.namprd04.prod.outlook.com]
> Queued mail for delivery)
the "a + b" delays here sum to 0.07 seconds. This is not a delayed
message, you're wasting everyone's time with this.
> Mar 31 09:11:18 smtp2 postfix/smtp[12171]: A594712400EF:
> to=<ab...@itssonline.com>, relay=smtp.secureserver.net[68.178.213.37]:25,
> delay=14, delays=0.05/0/5.6/8.3, dsn=5.2.0, status=bounced (host
> smtp.secureserver.net[68.178.213.37] said: 552 5.2.0 kDB41n0172ClyZo01 -
> kDB41n0172ClyZo01DBAKn IB212 msg rejected as spam (in reply to end of DATA
> command))
Ditto, but now we see that some receving systems concur that the
messages are likely spam.
> Apr 1 09:38:57 smtp2 postfix/smtp[20003]: 944671240115:
> to=<ab...@thoughtmill.com>, relay=none, delay=6.1, delays=0.04/3.5/2.6/0,
> dsn=4.4.3, status=deferred (Host or domain name not found. Name service
> error for name=thoughtmill.com type=MX: Host not found, try again)
Mail to a bogus domain, still low total delay, ...
> Apr 1 09:38:58 smtp2 postfix/bounce[28306]: 944671240115: sender
> non-delivery notification: 2769A124008B
Premature bounce due to disabled retries preventing interoperation
with greylisting...
> Apr 1 09:38:56 smtp2 postfix/smtp[19719]: 6E35F1240105:
> to=<aks...@reliableassociates.com>,
> relay=mailstore1.secureserver.net[216.69.186.201]:25, delay=31,
> delays=0.22/0.48/29/1.6, dsn=5.0.0, status=bounced (host
> mailstore1.secureserver.net[216.69.186.201] said: 550 #5.1.0 Address
> rejected. (in reply to RCPT TO command))
Poor list management still low cumulative delay, but finally a very
high $c, connecting to the backup MX.
> Apr 1 09:38:57 smtp2 postfix/smtp[19972]: 7A0FD1240142:
> to=<ab...@betheltech.net>, relay=none, delay=5.4, delays=0.05/5.3/0.02/0,
> dsn=5.4.4, status=bounced (Host or domain name not found. Name service error
> for name=betheltech.net type=A: Host not found)
> Apr 1 09:38:58 smtp2 postfix/bounce[28309]: 7A0FD1240142: sender
> non-delivery notification: 27AB912400CC
Still low $a + $b (the queue is not congested yet) and negligible
$c + $d) since destination domain is bogus and another bounce.
> The above are some of the results at the start of the day on 31st Mar and on
> 1st Apr. Let me know if you need any further information in this regard.
I'd done here. You're on your own.
--
Viktor.
