Need some help with Mailman and Sendmail
I had Mailman working, then went and updated to RH 6.1 which added the secure sendmail which I didn't know it would do. This immediately broke Mailman. After number recompiles, I gave up and reinstalled RH 6.1 from scratch. I have followed all the instructions in INSTALL.
Here is my aliases:
test: "|/etc/smrsh/wrapper post test" test-admin: "|/etc/smrsh/wrapper mailowner test" test-request: "|/etc/smrsh/wrapper mailcmd test" test-owner: test-admin
mailman: my email address mailman-owner: mailman
I have linked /etc/smrsh/wrapper /home/mailman/mail/wrapper
Here is the error I am getting:
The original message was received at Tue, 25 Jul 2000 11:28:16 -0500 from localhost.localdomain [127.0.0.1]
----- The following addresses had permanent fatal errors ----- <test-request@unix.ltlb.com>
----- Transcript of session follows ----- 553 unix.ltlb.com. config error: mail loops back to me (MX problem?) 554 <test-request@unix.ltlb.com>... Local configuration error
I didn't get any ./configure error nor any during make install. The first time I ran check-perms it found 0 errors. I did the make install as mailman with the source tar ball in a /home/mailman/test directory.
If more info is needed let me know. I've been beating my head against this for a week now.
Thanks,
Matt
On Tue, 25 Jul 2000, Matt Starnes wrote:
Hmm. You definitely need a symbolic link to the wrapper in/etc/smrsh, but you don't want to point your aliases at that synlink.
Instead, point the aliases directly to where the *real* wrapper is
located. Like this:
---BEGIN aliases.txt--- test: "|/home/of/mailman/wrapper post test" test-admin: "|/home/of/mailman/wrapper mailowner test" test-request: "|/home/of/mailman/wrapper mailcmd test" test-owner: test-admin ---END---
Try that and see what happens.
~ Rick ~-- .oooO "Man with closed Oooo. Rick C. Niess ( ) mouth gathers ( ) University of Southern Miss. \ ( no foot!" ) / resnet@usm.edu --\ )------------------(_/-------------------------------
Quoting Rick Niess (rniess@bark.cc.usm.edu):
Both ways work, at least they did when I had majordomo and mailman up simultaneously and I needed to disambiguate the two wrapper programs. I had two symlinks in /etc/smrh, one called majordomo_wrapper and the otehr called mailman_wrapper, and I referenced the links in /etc/smrsh from /etc/aliases.
-- Paul Tomblin <ptomblin@xcski.com>, not speaking for anybody There are three kinds of people: Those who can count & those who can't.
On Tue, 25 Jul 2000, Paul Tomblin wrote:
Ack, that leaves me stumped then. Have you checked your/var/log/maillog to see what errors are appearing there?
~ Rick ~-- .oooO "Man with closed Oooo. Rick C. Niess ( ) mouth gathers ( ) University of Southern Miss. \ ( no foot!" ) / resnet@usm.edu --\ )------------------(_/-------------------------------
On Tue, 25 Jul 2000, Matt Starnes wrote:
As others have mentioned, this is really the backwards way of doing things, and will likely confuse other administrators, but is not inherently a problem.
This is entirely a Sendmail misconfiguration problem. I noticed your mail came from @mail.ltlb.com, while this is coming from @unix.ltlb.com. From poking around, I notice you only have MX records for your domain (ltlb.com), not either of these hosts (mail.ltlb.com or unix.ltlb.com). If you have Sendmail configured to do masquerading properly this may not be a problem, but it is a good idea to have MX records for all hosts which mail may appears as addressed from.
Secondly, and this is where I think the problem lies, from what I gather from this error message and poking around in your DNS, unix.ltlb.com should connect to mail.ltlb.com to deliver it's mail, correct? Is unix.ltlb.com in mail.ltlb.com's /etc/mail/local-host-names? This is the Sendmail's replacement (in versions >8.9) for what used to be /etc/sendmail.cw; it is the list of hosts for whom you can accept mail. If unix.ltlb.com is not listed here, mail.ltlb.com will not accept mail for it, at which point it will bounce back to unix.ltlb.com, thus looping as the error message reports.
In addition, the O'Reilly DNS & BIND book points out another possible misconfiguration here: you should always use the canonical name of the host as the MX record. However, I can't really tell if you are in fact doing this, because your DNS seems to have just flapped. :( I also reasonably sure that you can get around this limitation with /etc/mail/local-host-names.
I hope these suggestions are helpful. The rule I always go by when trying to debug a Sendmail problem is "Have you checked /etc/mail/local-host-names"? It's very often the cause of your troubles.
-- Virginia J. Beauregard virginia@texterity.com UNIX Systems and Network Administrator Texterity, Inc.
You went past me with the MX records. I've heard of them, but I'm not totally sure whay they are. Basically we have two separate SMTP/POP3 servers, one running NT with Netscape software that is mail.ltlb.com, and one that is sendmail/qpopper on RH Linux 6.1. Emails to ltlb.com or mail.ltlb.com are handled by the NT box. I had Mailman working before until I updated to RH 6.1 which installed Sendmail with SMRSH "feature" which immediately broke many programs we have that route through sendmail. And the unix.ltlb.com box can send and receive standard email no problem. I think it is a smrsh problem, but I can't nail it down.
Matt
Virginia Beauregard wrote:
On Tue, 25 Jul 2000, Matt Starnes wrote:
Hmm. You definitely need a symbolic link to the wrapper in/etc/smrsh, but you don't want to point your aliases at that synlink.
Instead, point the aliases directly to where the *real* wrapper is
located. Like this:
---BEGIN aliases.txt--- test: "|/home/of/mailman/wrapper post test" test-admin: "|/home/of/mailman/wrapper mailowner test" test-request: "|/home/of/mailman/wrapper mailcmd test" test-owner: test-admin ---END---
Try that and see what happens.
~ Rick ~-- .oooO "Man with closed Oooo. Rick C. Niess ( ) mouth gathers ( ) University of Southern Miss. \ ( no foot!" ) / resnet@usm.edu --\ )------------------(_/-------------------------------
Quoting Rick Niess (rniess@bark.cc.usm.edu):
Both ways work, at least they did when I had majordomo and mailman up simultaneously and I needed to disambiguate the two wrapper programs. I had two symlinks in /etc/smrh, one called majordomo_wrapper and the otehr called mailman_wrapper, and I referenced the links in /etc/smrsh from /etc/aliases.
-- Paul Tomblin <ptomblin@xcski.com>, not speaking for anybody There are three kinds of people: Those who can count & those who can't.
On Tue, 25 Jul 2000, Paul Tomblin wrote:
Ack, that leaves me stumped then. Have you checked your/var/log/maillog to see what errors are appearing there?
~ Rick ~-- .oooO "Man with closed Oooo. Rick C. Niess ( ) mouth gathers ( ) University of Southern Miss. \ ( no foot!" ) / resnet@usm.edu --\ )------------------(_/-------------------------------
On Tue, 25 Jul 2000, Matt Starnes wrote:
As others have mentioned, this is really the backwards way of doing things, and will likely confuse other administrators, but is not inherently a problem.
This is entirely a Sendmail misconfiguration problem. I noticed your mail came from @mail.ltlb.com, while this is coming from @unix.ltlb.com. From poking around, I notice you only have MX records for your domain (ltlb.com), not either of these hosts (mail.ltlb.com or unix.ltlb.com). If you have Sendmail configured to do masquerading properly this may not be a problem, but it is a good idea to have MX records for all hosts which mail may appears as addressed from.
Secondly, and this is where I think the problem lies, from what I gather from this error message and poking around in your DNS, unix.ltlb.com should connect to mail.ltlb.com to deliver it's mail, correct? Is unix.ltlb.com in mail.ltlb.com's /etc/mail/local-host-names? This is the Sendmail's replacement (in versions >8.9) for what used to be /etc/sendmail.cw; it is the list of hosts for whom you can accept mail. If unix.ltlb.com is not listed here, mail.ltlb.com will not accept mail for it, at which point it will bounce back to unix.ltlb.com, thus looping as the error message reports.
In addition, the O'Reilly DNS & BIND book points out another possible misconfiguration here: you should always use the canonical name of the host as the MX record. However, I can't really tell if you are in fact doing this, because your DNS seems to have just flapped. :( I also reasonably sure that you can get around this limitation with /etc/mail/local-host-names.
I hope these suggestions are helpful. The rule I always go by when trying to debug a Sendmail problem is "Have you checked /etc/mail/local-host-names"? It's very often the cause of your troubles.
-- Virginia J. Beauregard virginia@texterity.com UNIX Systems and Network Administrator Texterity, Inc.
You went past me with the MX records. I've heard of them, but I'm not totally sure whay they are. Basically we have two separate SMTP/POP3 servers, one running NT with Netscape software that is mail.ltlb.com, and one that is sendmail/qpopper on RH Linux 6.1. Emails to ltlb.com or mail.ltlb.com are handled by the NT box. I had Mailman working before until I updated to RH 6.1 which installed Sendmail with SMRSH "feature" which immediately broke many programs we have that route through sendmail. And the unix.ltlb.com box can send and receive standard email no problem. I think it is a smrsh problem, but I can't nail it down.
Matt
Virginia Beauregard wrote:
participants (4)
-
Matt Starnes -
Paul Tomblin -
Rick Niess -
Virginia Beauregard