Sendmail not working

[Andrea Pearsall]

Ok. Here goes....

I have two rp2470 servers. Each server runs identical copies of sendmail version 8.9.3. Each one has an identical sendmail.cf file. The servers are used for Oracle Financial applications and only I am only concerned with sending email out of the server. To do this I have configured the sendmail.cf to have a DS macro of the server that will relay the mail. I have shutdown the sendmail service for both server and disabled it from automatically starting at boot.

Here is the problem. I can easily recieve emails from one server but I never get eamils from the other. The relay server is a Microsoft exchange server version 6.

When I use the "-v" flag of sendmail I get identical output for both with the exception of a few bits that should be different like message id and stuff.

The two servers are in a domain that I am not in as we are a management company. Users within the domain of the servers get emails from the server that I cannot. I'm perplexed.

Both servers were working before I applied the september '03 quality pack. I have done an "swlist -l product | grep PH" on both servers and compared. There were about 7 patches difference but there was an older ARPA patch on the server that does not send mails to my email address.

I have talked to support and they say the problem is with the exchange server and not with HP but I know little about Exchange and the people that run Exchange at the company we manage don't seem to up on the play.

I just wan't to find out where the email is getting stuck for one but not the other...

Any ideas???

[Patrick Wallek · Accepted solution]

What does /var/adm/syslog/mail.log show on the server that you don't get messages from?

Try to send a message and post the relevant output from the mail.log. Maybe that'll give us some clues.

Also make sure that you can resolve the IP address of the exchange server via nslookup. Try nslookup with both the name and IP address and make sure they both work. You could also try adding the name/ip of the exchange server to the /etc/hosts file and see if that helps.

[Andrea Pearsall]

Thanks Patrick,

For some reason the /var/adm/syslog/mail.log is always empty, even if I (successfully?) send an email. I can resolve the exchange server by name and IP, as well I have the fully qualified name of the echange server in the hosts file. My resolution is FILES -> DNS.

I find it strange that the /var/adm/syslog/mail.log does not show any messages for the meail being sent. I recently cleared out the mail.log by moving it to /tmp and recreating it. I thought there might be a problem with doing that if the process has a lock on the inode of the mail.log file.

Im gonna give you full points though for helping my pull my head out of my --- because I never thought to check to see if the server that was sending mail was actually putting an entry in the log (which it is as one would expect). I now have a lead! Why isn't mail.log logging??? This is at least a clue.

Henri

[Patrick Wallek]

OK, a couple of things to check here.

1) Check /etc/syslogd.conf and make sure mail messages are configured to go to the mail.log file.

2) If you did move the file and recreate while syslogd was running then you probably just need to stop and restart the syslogd daemon.

# /sbin/init.d/syslogd stop
# /sbin/init.d/syslogd start

Then try sending a message again and see if anything gets logged.

[Steven E. Protter]

Regardless of the logging issue, it should be quite easy to see whats going on with the mail, even without the mail.log file.

As root

sendmail -v -d7.99 -d38.99 root@secondmachin
type some text then hit enter

.



You will get so much diagnostic information you might want to redirect it to a file. The problem will be clear as to what is going on with the mail.

Then you can post and take corrective action.

SEP

[Andrea Pearsall]

Thanks for the input. I did restart the syslogd and now logging is re-enabled. I checked the log and the mail seems to be leaving the server via an smtp connection to the exchange server as expected. I compared the output of the full debug command of both servers and they are identical in all the important areas. I am now getting a head of steam to go after the exchange administrators to see where the mail is going. I checked the mailq on my server and nothing is there. There are no dead.letters.

If any of you know any good questions to as them then by all means....They use Mail Marshall and Exchange version 6. I use them as a very simple relay.

Cheers.

[Patrick Wallek]

I would make sure that they are allowing you to relay. Have them verify that both of your servers are allowed to submit mail for it to relay. It sounds as if one is allowed and one is not, and for the on that is not they are just dropping the messages internally on the exchange server.

[Berlene Herren]

Try running a debug...

#/usr/sbin/sendmail -bt
ADDRESS TEST MODE (ruleset 3 NOT automatically invoked)
Enter

> .D{client_name} sid.atl.hp.com
> .D{client_addr} 15.17.184.8
> check_rcpt user@hostname.com


The address of the client attempting the relay.
.D{client_name} myclient.mydomain
.D{client_addr} xxx.xxx.xxx.xxx


Berlene