Multiple IMC's, delivery failures/NDR's

Multiple IMC's, delivery failures/NDR's

Post by Kamal Jai » Fri, 04 May 2001 23:53:24



Hello,

Every now and then some email destined for the Internet bounces with this
error:

The following recipient(s) could not be reached:

     Unable to deliver the message due to a communications failure
        The MTS-ID of the original message is: c=US;a=
;p=Marathon;l=ZACH-010501112452Z-511
      MSEXCH:IMS:Marathon:HQ:ZACH 0x80040C02 (00050311) Maximum receipt
headers exceeded (possible mail loop).

It's so intermittent that I'm having trouble figuring it out.  We have 1
site, 1 organization, and 4 servers.  All servers are Exchange 5.5sp4
running on NT 4.0sp6a, 3 of the servers send and receive email with the
Internet through firewalls, and one of them (the one I'm on) is running the
IMC for internal processes that use CDONTS to send email.  This last server
has IMC setup to route all email through one of the other servers.

Can anyone recommend a good reference to explain best/recommended
configurations for IMC's?  It all seems simple enough, but I somehow suspect
that I'm missing something.

Any help will be appreciated.  To email me, remove the tildes (~) from my
address, otherwise just post here.

Thanks,
- kamal
--
Kamal Jain, IT/Facilities Manager
Marathon Technologies Corporation
tel. +1 978 489 1123  fax. +1 978 489 1099

http://www.marathontechnologies.com
Marathon Assured Availability (tm) Solutions for Microsoft Windows
environments

 
 
 

Multiple IMC's, delivery failures/NDR's

Post by Rich Matheisen [MVP » Sat, 05 May 2001 10:16:37



>Hello,

>Every now and then some email destined for the Internet bounces with this
>error:

>The following recipient(s) could not be reached:

>     Unable to deliver the message due to a communications failure
>        The MTS-ID of the original message is: c=US;a=
>;p=Marathon;l=ZACH-010501112452Z-511
>      MSEXCH:IMS:Marathon:HQ:ZACH 0x80040C02 (00050311) Maximum receipt
>headers exceeded (possible mail loop).

It simply means that the message is looping between your server and
the destination server.

If this is happening for outbond mail, and you use DNS instead of a
smarthost, it's probably a bad MX or A record in DNS. Your server may
be sending the mail to itself.

If it's happening on inbound mail it's usually a problem with ther
contents of the IMS's "Routing" tab. There's a MX record that points
to your server, but your server doesn't accept the mail as <inbound>.

Quote:>It's so intermittent that I'm having trouble figuring it out.  We have 1
>site, 1 organization, and 4 servers.  All servers are Exchange 5.5sp4
>running on NT 4.0sp6a, 3 of the servers send and receive email with the
>Internet through firewalls, and one of them (the one I'm on) is running the
>IMC for internal processes that use CDONTS to send email.  This last server
>has IMC setup to route all email through one of the other servers.

Why so many IMS's?

Quote:>Can anyone recommend a good reference to explain best/recommended
>configurations for IMC's?  It all seems simple enough, but I somehow suspect
>that I'm missing something.

Unless those IMS's are widely dispersed, you could probably get away
with only one. The IMS in Exchange 5.5 offers no redundancy, so having
more than one isn't very useful (there are exceptions, of course).

------------------
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm

 
 
 

Multiple IMC's, delivery failures/NDR's

Post by Kamal Jai » Sat, 05 May 2001 22:38:46


Thanks for the response, Rich.

One of the servers is in another country, connected to the home office via
VPN, but they have their own email domain and Internet connection there, so
there's one IMC.  Here in the home office I have two IMC's connected to the
Internet for inbound redundancy (two MX records with different costs).  And
the other one here is, as I said, for CDONTS.

We are accepting inbound SMTP mail for 4 different email domains onto all
the servers, once again, to provide some redundancy with inbound email.

[Of course, from a practical perspective it seems many mailers out on the
Internet do not try multiple MX hosts -- they try the first MX host they
find and then give up if it's not available.]

All the servers except the CDONTS internal server are set to use DNS, but
the email was always destined for an email domain that our DNS servers don't
have zone files for.  I am wondering about possible DNS cache corruption,
but that's pretty unusual.

It also sometimes seems like maybe there is something in the email message
that's causing trouble.  If my brain serves me this morning, in all cases,
it has been a reply to an email that fails, even if I delete all the
recipients and re-enter by hand.  But a new email to them, with the text
copied from the original message, seems to go through fine.

Has anyone seen this behavior?

- kamal
--
Kamal Jain, IT/Facilities Manager
Marathon Technologies Corporation
tel. +1 978 489 1123  fax. +1 978 489 1099

http://www.marathontechnologies.com
Marathon Assured Availability (tm) Solutions for Microsoft Windows
environments

 
 
 

Multiple IMC's, delivery failures/NDR's

Post by Rich Matheisen [MVP » Sun, 06 May 2001 09:28:35



>One of the servers is in another country, connected to the home office via
>VPN, but they have their own email domain and Internet connection there, so
>there's one IMC.  Here in the home office I have two IMC's connected to the
>Internet for inbound redundancy (two MX records with different costs).  And
>the other one here is, as I said, for CDONTS.

Okay. That accounts for 'em all!

Quote:>We are accepting inbound SMTP mail for 4 different email domains onto all
>the servers, once again, to provide some redundancy with inbound email.

That's not a problem, provided you don't prefix the domain with a "#"
in the IMS "Routing" tab. If you do that the IMS accepts only mail
that matches the domain /exactly/ -- no subdomains will be accepted as
<inbound>.

Quote:>[Of course, from a practical perspective it seems many mailers out on the
>Internet do not try multiple MX hosts -- they try the first MX host they
>find and then give up if it's not available.]

That's their loss. :)

Quote:>All the servers except the CDONTS internal server are set to use DNS, but
>the email was always destined for an email domain that our DNS servers don't
>have zone files for.  

Well, I *hope* not! You couldn't be authoritative for every domain on
the Internet!

Quote:>I am wondering about possible DNS cache corruption,
>but that's pretty unusual.

Ummm, not really. A bad answer to a query, or malicious poisoning of
DNS are pretty common. If the bad answer has a TTL of several years .
. . well, you can see what a problem that might cause. :) Oh, and to
add to that, the IMS also caches the results of name queries. If you
suspect a DNS problem, flush the DNS cache and then stop and restart
the IMS to flush its cache, too.

Quote:>It also sometimes seems like maybe there is something in the email message
>that's causing trouble.  

Not likely. The IMS doesn't use the contents of the message to iscover
where it should go. It uses only the RCPT TO commands from the
transmitting SMTP server.

Quote:>If my brain serves me this morning, in all cases,
>it has been a reply to an email that fails, even if I delete all the
>recipients and re-enter by hand.  But a new email to them, with the text
>copied from the original message, seems to go through fine.

The "Received:" headers in the message will tell you what's happening
pretty quickly.

------------------
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm