sendmail problem on 5.0.4 on SW Bell ADSL line
Steve Fabac
that begin complaining about long delivery times for their outgoing
e-mail. They reported a 5 hour delay between the time they sent the
email and the recipient receiving it. (No, I have not asked the client
to have the recipient forward me the email so that I can check the
headers ... yet.)
To try to avoid the seeming delay in the swbell mail server, I commented
out the "smart host" directive in sendmail.cf that specified
"mail.swbell.net"
Mail to smf...@swbell.net then fails with
> Mail Queue (1 request)
> --Q-ID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------
> CAA29561 34 Fri Jun 1 02:55 smf
> (smf...@swbell.net... reply: read error from swbell.net.)
> smf...@swbell.net
But if I send the message addressed to smf...@mail.swbell.net
it gets off the machine and is delivered!!!
The following is a test message with headers when sent with smart
mail host in sendmail.cf set to mail.swbell.net:
> Return-path:<s...@insSPAMlass.com>
> Received:from mta5.rcsntx.swbell.net (mta5-pr.rcsntx.swbell.net)
> by sims2.rcsntx.swbell.net (Sun Internet
> Mail Server sims.3.5.2000.01.05.12.18.p9) with ESMTP id
> <0GE800C...@sims2.rcsntx.swbell.net>
> for smfabac@sims-ms-daemon; Fri, 1 Jun 2001 02:17:26 -0500 (CDT)
> Received: from unx2.insSPAMlass.com ([64.219.XXX.XXX]) by
> mta5.rcsntx.swbell.net (Sun Internet Mail Server
> sims.3.5.2000.03.23.18.03.p10) with ESMTP id > <0GE800E...@mta5.rcsntx.swbell.net>
> for smf...@sims2.rcsntx.swbell.net; Fri, 01 Jun 2001 02:14:44 -0500 (CDT)
> Received: (from smf@localhost) by unx2.insSPAMlass.com (8.8.8/SCO5)
> id BAA28238 for smf...@swbell.net; Fri, 01 Jun 2001 01:58:22 -0500 (CDT)
> Date: Fri, 01 Jun 2001 01:58:22 -0500 (CDT)
> From: s...@insSPAMlass.com (Steve Fabac)
> Subject: testing e-mail
> To: smf...@swbell.net
> Message-ID: <2001060106...@unx2.insSPAMlass.com>
> X-Mailer: SCO OpenServer Mail Release 5.0
> X-Mozilla-Status: 0000
>
> This is being sent to smf...@swbell.net
This next message is without the smart host and the message was
sent to smf...@mail.swbell.net. Note that the "to" line has been
re-written as smf...@swbell.net by the SWBELL mail system.
> Return-path: <s...@insSPAMlass.com>
> Received: from mta4.rcsntx.swbell.net (mta4-pr.rcsntx.swbell.net)
> by sims2.rcsntx.swbell.net (Sun Internet
> Mail Server sims.3.5.2000.01.05.12.18.p9) with ESMTP id
> <0GE800F...@sims2.rcsntx.swbell.net> for
> smfabac@sims-ms-daemon; Fri, 1 Jun 2001 02:54:42 -0500 (CDT)
> Received: from unx2.insSPAMlass.com ([64.219.142.113]) by
> mta4.rcsntx.swbell.net (Sun Internet Mail Server
> sims.3.5.2000.03.23.18.03.p10) with ESMTP id > <0GE8000...@mta4.rcsntx.swbell.net> for
> smf...@sims2.rcsntx.swbell.net; Fri, 01 Jun 2001 02:54:51 -0500 (CDT)
> Received: (from root@localhost) by unx2.insSPAMlass.com (8.8.8/SCO5)
> id CAA29561 for smf...@swbell.net; Fri, 01 Jun 2001 02:55:26 -0500 (CDT)
> Date: Fri, 01 Jun 2001 02:55:26 -0500 (CDT)
> From: ro...@insSPAMlass.com (Superuser)
> Subject: testing e-mail without smarthost
>>> To: smf...@swbell.net
> Message-ID: <2001060107...@unx2.insSPAMlass.com>
> X-Mailer: SCO OpenServer Mail Release 5.0
> X-Mozilla-Status: 8001
>
>
> This is a test without smarthost
>
Can anyone shed some light on this problem? Is there a known problem
between SCO and the Sun Internet Mail Server
(sims.3.5.2000.03.23.18.03.p10) being used by SW BELL?
Steve Fabac
S.M. Fabac & Associates
816/765-1670
Bill Vermillion
>I have a client running SCO 5.0.4 Ent. with sse022 (sendmail 8.8.8)
>that begin complaining about long delivery times for their outgoing
>e-mail. They reported a 5 hour delay between the time they sent the
>email and the recipient receiving it. (No, I have not asked the client
>to have the recipient forward me the email so that I can check the
>headers ... yet.)
>To try to avoid the seeming delay in the swbell mail server, I
>commented out the "smart host" directive in sendmail.cf that
>specified "mail.swbell.net"
>Mail to smf...@swbell.net then fails with
>> Mail Queue (1 request)
>> --Q-ID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------
>> CAA29561 34 Fri Jun 1 02:55 smf
>> (smf...@swbell.net... reply: read error from swbell.net.)
>> smf...@swbell.net
>But if I send the message addressed to smf...@mail.swbell.net
>it gets off the machine and is delivered!!!
No need for exclamation points there, as that is exactly the
expected behaviour if you so not have SmartHost and DO NOT have
a local DNS server.
By commenting out 'SmartHost' - which will lookup the
proper MX records, you now have to deliver mail to an explicit
email address, as you have just found. That's why it's
called a "SmartHost".
>Can anyone shed some light on this problem? Is there a known problem
>between SCO and the Sun Internet Mail Server
>(sims.3.5.2000.03.23.18.03.p10) being used by SW BELL?
Shouldn't be. But you made changes based on slow mail delivery.
Unless you get the email you will never know why, so I'd advise not
making changes until you find the problem.
I had mail delayed for three days - I was acting as a tertiary MX
server for a local .org group, and fibre cuts made both their
primary and secondary unseeable.
RoadRunner mail in Texas was backed up for 3 or 4 days a couple of
weeks ago - as evidenced by a friend who has an account there and I
was getting all the undeliverables.
A five hour delay is normal IF the connection can't be made rather
quickly as that could be the retry time [my severs are set for four
hours].
I'd not worry about the occaisional mail delay. It's a fact of
life in todays 'net.
--
Bill Vermillion - bv @ wjv . com
Steve Fabac
> Bill Vermillion - bv @ wjv . com
You're right Bill, I found that sendmail is not using DNS on the
client's machine due to the existence of /etc/service.switch.
Some time in the past before the client was connected to the internet,
I had followed the SCO technical article to create /etc/service.switch
to prevent sendmail from using the nonexistent DNS service for
intra-company e-mail. When the system was connected to the internet, I
reconfigured /usr/lib/sendmail.cf and dropped the OI-DNSRCH but
did not remove /etc/service.switch.
Well, I fought getting e-mail off the machine until I set
mail.swbell.net
as the "smart host." Then, all non local e-mail was relayed through
swbell.net.
I found this when I executed /usr/lib/sendmail -bt and
issued /mx swbell.net and received the following response
> With /etc/service.switch set to:
> hosts files /dns
>
> # /usr/lib/sendmail -bt
> WARNING: writable directory /usr
> WARNING: writable directory /usr/spool
> WARNING: writable directory /usr/spool/mqueue
> ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
> Enter <ruleset> <address>
> /mx swbell.net
> > getmxrr(swbell.net) returns 0 value(s):
> >
>
> After removing service.switch or changing it to:
> hosts files dns
>
> # /usr/lib/sendmail -bt
> WARNING: writable directory /usr
> WARNING: writable directory /usr/spool
> WARNING: writable directory /usr/spool/mqueue
> ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
> Enter <ruleset> <address>
> /mx swbell.net
> > getmxrr(swbell.net) returns 3 value(s):
> mta1.rcsntx.swbell.net.
> mta2.rcsntx.swbell.net.
> mta3.rcsntx.swbell.net.
> >
Problem solved.
Bill Vermillion
>> In article <3B1459D9...@att.net>, Steve Fabac <smf...@att.net> wrote:
>> >I have a client running SCO 5.0.4 Ent. with sse022 (sendmail 8.8.8)
>> >that begin complaining about long delivery times for their outgoing
>> >e-mail. ...
...
>> >Mail to smf...@swbell.net then fails with
>>
>> >> Mail Queue (1 request)
>> >> --Q-ID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------
>> >> CAA29561 34 Fri Jun 1 02:55 smf
>> >> (smf...@swbell.net... reply: read error from swbell.net.)
>> >> smf...@swbell.net
>> >But if I send the message addressed to smf...@mail.swbell.net
>> >it gets off the machine and is delivered!!!
>> No need for exclamation points there, as that is exactly the
>> expected behaviour if you so not have SmartHost and DO NOT have
>> a local DNS server.
>You're right Bill, I found that sendmail is not using DNS on the
>client's machine due to the existence of /etc/service.switch.
I've had days like that. Just had one after getting power back
after 12.5 hours. Tree fell across lines and snapped a powerpole.
You can't do much without electricity anymore.
>Some time in the past before the client was connected to the internet,
>I had followed the SCO technical article to create /etc/service.switch
>to prevent sendmail from using the nonexistent DNS service for
>intra-company e-mail. When the system was connected to the internet, I
>reconfigured /usr/lib/sendmail.cf and dropped the OI-DNSRCH but
>did not remove /etc/service.switch.
>> With /etc/service.switch set to:
>> hosts files /dns
>> # /usr/lib/sendmail -bt
>> WARNING: writable directory /usr
>> WARNING: writable directory /usr/spool
>> WARNING: writable directory /usr/spool/mqueue
>> ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
>> Enter <ruleset> <address>
>> /mx swbell.net
>> > getmxrr(swbell.net) returns 0 value(s):
>> After removing service.switch or changing it to:
>> hosts files dns
>>
>> # /usr/lib/sendmail -bt
>> WARNING: writable directory /usr
>> WARNING: writable directory /usr/spool
>> WARNING: writable directory /usr/spool/mqueue
>> ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
>> Enter <ruleset> <address>
>> /mx swbell.net
>> > getmxrr(swbell.net) returns 3 value(s):
>> mta1.rcsntx.swbell.net.
>> mta2.rcsntx.swbell.net.
>> mta3.rcsntx.swbell.net.
>Problem solved.
Glad it's working. I do enough DNS and sendmail that it looked
like that was the problem.