[SOLVED] NDRs (2.0.0) for too many hops are being sent to sendmail's postmaster

DennisC31 · 12-05-2011, 11:39 AM

So... sendmail gurus. Help!

I have finally gotten fed up with this problem:

1. User Dennis sets up his out of office reply and leaves for a vacation.
2. Server PonyExpress sends dennis a routine report on something.
3. The mail router mrouter1 gets the message from PonyExpress and forwards it to Dennis on the Exchange server.
4. Exchange sends a reply from dennis to root@ponyexpress.company.com to tell root that dennis is on vacation.
5. mrouter1 gets the message for root@ponyexpress.company.com but it doesn't know who to send it to so it sends it to the external mail server for processing.
6. The external mail server sees it as something for company.com so it sends it back to mrouter1.

(5 and 6 repeat until 26 hops)

7. I get these notices in the maillog on mrouter1:

Dec 5 12:03:42 mrouter1 sendmail[15139]: pB5GxWQn014254: pB5H3g9w015139: postmaster notify: Too many hops 26 (25 max): from <dennis@company.com> via localhost, to <root@ponyexpress.company.com>
Dec 5 12:03:42 mrouter1 sendmail[15139]: pB5H3g9w015139: to=/dev/null, ctladdr=dev-null (8/0), delay=00:00:00, xdelay=00:00:00, mailer=*file*, pri=90000, dsn=2.0.0, stat=Sent
Dec 5 12:03:43 mrouter1 sendmail[15139]: pB5H3g9w015139: to=dennisboss@company.com,dennis@company.com, delay=00:00:01, xdelay=00:00:01, mailer=esmtp, pri=90000, relay=exchange01.corp.company.com. [10.15.61.68], dsn=2.0.0, stat=Sent (<201112051703.pB5H3g9w015139@mrouter1.company.com> Queued mail for delivery)

Notice that it correctly sends the DSN to null but then it sends one to me and my boss -- and we are both aliased as sa@company.com so presumably somewhere... something is set to send to sa@company.com.

The closest thing I can find is for a double bounce being sent this way. But my address IS valid so not sure if thats it. I have tried various combinations of these settings in sendmail.mc but to no avail -- same behaviour each time.

What could possibly be doing this?

I am pulling my hair out.

Maybe the magic muse will make the light come on 5-10 minutes after I hit submit.

bathory · 12-05-2011, 01:27 PM

Quote:

Notice that it correctly sends the DSN to null but then it sends one to me and my boss -- and we are both aliased as sa@company.com so presumably somewhere... something is set to send to sa@company.com.

Check /etc/aliases (or /etc/mail/aliases depending on distro), to see if postmaster is aliased to sa@company.com

DennisC31 · 12-06-2011, 08:19 AM

K. We have a solution.

Postmaster is set to sa@company.com and sa@company.com goes to Dennis and DennisBoss@company.com.

I did some reading on release notes of sendmail and they actually have said:

Code:

Make sure that "too many hops" messages are considered important
enough to send an error to the Postmaster (that is, the
address specified in the P option). --release notes for 8.6.6/8.6.6 1994/03/13

It seems that there is no way to turn off these postmaster notifies for too many hops (2.0.0) errors. I can understand why. Too many hops usually indicate a seriously broken mail route. In my case, its not a big deal, we want these messages to die but we don't want the postmaster notify spam.

The fix, although not elegant, is to add a bogus entry into the mailertable for the offending domain.

So I added:

Code:

offendingserver.company.com esmtp:unresolvableaddress.company.com

to /etc/mail/mailertable and then reloaded Sendmail. Now when a message goes out to someone@offendingserver.company.com, the sender gets an NDR because the domain name is not resolvable. Which is exactly what we want.

Again its not elegant, and there should be a way to control these in Sendmail (maybe "dontblamesendmailpart2"). However I can live with the result above.