Sendmail - "host map: lookup (example.com): deferred)" problem

[Vitaly Karasik_1]

I have RHEL4 with sendmail-8.13.1-3.RHEL4.5. I didn't change sendmail config.
Many times I see this error when I send email from this host to outside:

#mailq
/var/spool/mqueue (1 request)
-----Q-ID----- --Size-- -----Q-Time----- ------------Sender/Recipient-----------
kB7ASdUr001133 0 Thu Dec 7 10:28
(host map: lookup (example.com): deferred)

Total requests: 1

more details:
1) I get this error random for all domains I try
2) IMHO, there is no DNS problem - "host -t mx example.com" provides me an answer in 0.0000...1 sec
3) weird thing - in maillog I don't see messages related to this problem even I have loglevel = 9
4)after some time or several "sendmail -q" email goes to recipient

[Alpha977]

Hello Vitaly!

This server use another remailer or is public on internet?

[Vitaly Karasik_1]

"I didn't change sendmail config" - I mean, I don't use smart (or other) relay. This server uses "internet" DNS for routing emails.

[Vernon Brown_4]

Last couple of weeks have seen heavy DoS attacks on nameservers. Everydns.net was shut down last week end. So if this were a contributor to the problem you would see intermittent functionality.

Happened to me this weekend because of the Everydns denial of service attack.

[Vitaly Karasik_1]

of course, it's not 100% proof of my concept, but I run "host -t example.com" just after I see my email stucked in the queue... and it works very-very fast!

[Steven E. Protter]

Shalom Vitaly,

Here in Southern California today. Greetings and such.

I'm wondering if the problem is restricted to certain domains and not so random. If so, ths could be caused by sendmail configuration at the other end of the connection. e.g. that server is busy and is telling your machine to wait based on conditions on the receiving server. That's a condition I believe I could manipulate a sendmail server into doing with certain anti-spam or limits on how many open sendmail connections I permit.

I believe this to be an issue with either the receiving server or perhaps some kind of interference from your ISP/NSP deferring or limiting the use of port 25 smtp connections. The latter worth looking into.

As you know the default sendmail log level is 16 and you may need to increase that to get it to show up on the maillog file.

I agree with you on 1, though an issue with the DNS server in /etc/resolv.conf may be causing looking delays. The recipient domain could be delaying or defferring some DNS requests.

In summary, I think its something with the recieving server and not your system, though it could be a sendmail bug.

Lets say you send the mail to my address at NDS, does the message ever happen with that domain, which to my knowledge has a high threashold before deferring and delaying sendmail connections?

We're no longer using sendmail for inbound mail btw but do have traffic shaping and limiting features in place.

SEP

[Vitaly Karasik_1]

shalom from Jerusalem!

>I'm wondering if the problem is restricted >to certain domains and not so random. If
In fact, I tried two recipients - my work and my private email addresses. both sometimes work, sometimes not.

>I believe this to be an issue with either >the receiving server or perhaps some kind >of interference from your ISP/NSP >deferring or limiting the use of port 25 >smtp connections. The latter worth looking >into.
I'll try to find some RHEL3 machine located at the same hoster....
I have tens of RHEL3 machines located in another host farm, they doens't have this problem.


>As you know the default sendmail log level >is 16 and you may need to increase that to >get it to show up on the maillog file.
AFAIK, deafult is 9. anywhere - I'll increase it.