configuration glitch causing mail delivery fail
I recently moved my email server (with mailman) onto a new server with a different host name. In previous moves the server name was always email.marmot.org (which has an mx record). The server it's on now has a different name and does not have an mx record. I did add the old IP address to the public interface so DNS still points to that box as the mail server and by the name email.marmot.org and I have that in /etc/hosts.
Unfortunately when mailman sends messages they fail to deliver to most of the list subscribers and the maillog shows that the helo used is the actual name of the server, not the name of the email server. Naturally the receiving server disconnects when the names isn't right.
I could understand this with the move to the new server except every where I know to configure this (both postfix and mailman) I have the name email.marmot.org configured as the URL of the host. Regular email works fine from that box, just mailman is having a problem sending.
Does anyone have any idea what I'm missing?
Steve Lindemann, MSIS __ Network Administrator //\\ ASCII Ribbon Campaign Marmot Library Network \\// against HTML/RTF email, +1.970.242.3331 x116 //\\ vCards & M$ attachments http://www.marmot.org
On 01/07/2014 03:18 PM, Steve Lindemann wrote:
Unfortunately when mailman sends messages they fail to deliver to most of the list subscribers and the maillog shows that the helo used is the actual name of the server, not the name of the email server. Naturally the receiving server disconnects when the names isn't right.
How is Mailman delivering mail? If it is delivering to a local MTA on localhost, how it identifies itself in HELO to this MTA is irrelevant as it is this MTA, not Mailman that is attempting delivery to the recipient MX.
I could understand this with the move to the new server except every where I know to configure this (both postfix and mailman) I have the name email.marmot.org configured as the URL of the host. Regular email works fine from that box, just mailman is having a problem sending.
Does anyone have any idea what I'm missing?
In the normal case, Mailman relays mail through the local MTA (Postfix). How Mailman identifies itself to this MTA is irrelevant as long as the MTA accepts the mail from Mailman (does it)?
If the local postfix can deliver mail, but not Mailman's mail, the only thing I can see that would affect this is the envelope sender which would be LISTNAME-bounces@HOSTNAME where HOSTNAME is the list's host_name attribute.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 1/7/2014 4:40 PM, Mark Sapiro wrote:
On 01/07/2014 03:18 PM, Steve Lindemann wrote:
Unfortunately when mailman sends messages they fail to deliver to most of the list subscribers and the maillog shows that the helo used is the actual name of the server, not the name of the email server. Naturally the receiving server disconnects when the names isn't right.
How is Mailman delivering mail? If it is delivering to a local MTA on localhost, how it identifies itself in HELO to this MTA is irrelevant as it is this MTA, not Mailman that is attempting delivery to the recipient MX.
I could understand this with the move to the new server except every where I know to configure this (both postfix and mailman) I have the name email.marmot.org configured as the URL of the host. Regular email works fine from that box, just mailman is having a problem sending.
Does anyone have any idea what I'm missing?
In the normal case, Mailman relays mail through the local MTA (Postfix). How Mailman identifies itself to this MTA is irrelevant as long as the MTA accepts the mail from Mailman (does it)?
If the local postfix can deliver mail, but not Mailman's mail, the only thing I can see that would affect this is the envelope sender which would be LISTNAME-bounces@HOSTNAME where HOSTNAME is the list's host_name attribute.
Damn... that was fun. It took two things to get it working again. I needed to rename the server to something with a mx record, in this case: email.marmot.org vrs what it was. AND I had to turn off ipv6 on the interfaces. Mailman and postfix are on the same box. Somehow the mailman messages where picking up the ipv6 localhost address as the source of the original message and it appears that being rejected by receiving servers. Once I renamed the server and ditched ipv6 the messages started delivering again.
Steve
On 01/08/2014 01:43 PM, Steve Lindemann wrote:
Somehow the mailman messages where picking up the ipv6 localhost address as the source of the original message ...
Probably because the /etc/hosts entry for 'localhost' is the ipv6 address.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 1/8/2014 5:46 PM, Mark Sapiro wrote:
On 01/08/2014 01:43 PM, Steve Lindemann wrote:
Somehow the mailman messages where picking up the ipv6 localhost address as the source of the original message ...
Probably because the /etc/hosts entry for 'localhost' is the ipv6 address.
oh damn... I never even thought to look there. I really do need to remember to keep it simple and go with the basics. Especially since turning ipv6 off in centos turned out to be a lot harder than I ever expected it should be.
The ipv6 localhost address was the second line in the hosts file and ipv4 was the first. I thought the first line would have taken precedence unless mailman is specifically asking for the ipv6 localhost first.
Either way, I've cleaned that up now too. Thanks!
participants (2)
-
Mark Sapiro -
Steve Lindemann