Just started to set up email today on a fresh-out-of-the-box Centos 5.5 machine.
Af first, neither sending nor receiving mail would work. Browsing this forum, I found that the solution the receiving issue was changing /etc/dovecot/dovecot.conf maildir from: mail_location = maildir:~/Maildir:INDEX=/var/lib/dovecot/index/%u:CONTROL=/var/lib/dovecot/control/%u to mail_location = maildir:~/Maildir
However, I still can't send. The main client just a hangs. I've created email users on three virtual server, and the problems seems to be system-wide.
I can offer the following excerpt from /var/log/maillog which may mean something to someone:
Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: timeout after UNKNOWN from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: disconnect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21808]: timeout after UNKNOWN from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21808]: disconnect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: connect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21808]: connect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: lost connection after EHLO from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: disconnect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49] Feb 15 15:49:55 server88-208-236-226 postfix/smtpd[21806]: connect from host81-142-86-49.in-addr.btopenworld.com[81.142.86.49]
Also, this is the same machine where I had an issue with Virtualmin failing to pick up the system hostname (solved in a separate forum posting on here). Just thought I'd mention that in case the two may be related.
Howdy,
Those timeout errors are odd... typically, if a client can connect to Postfix at all, it wouldn't just timeout like that.
Do you by chance have another email client available to you, that you could see if you got similar results with it?
-Eric
Hi Eric,
Once again, thanks for your help.
I set up an account in Outlook 2010 which does indeed provide a more tangible error message upon trying to send "The server does not support secure connections".
The corresponding lines from the error log look like this:
Any thoughts? Isn't SSL enabled by default on Virtualmin?
Howdy,
Well, it sounds like it may be trying to use SSL on port 25, which wouldn't work.
Port 465 and 587 should support SSL -- but not by default :-)
The version of Postfix that comes with CentOS doesn't have SSL setup within , but you can do that.
The quickest way to do that would be to enable SSL for a domain in Virtualmin, then go into Server Configuration -> Manage SSL Certificates, and click "Copy to Postfix".
Then, edit /etc/postfix/master.cf, and uncomment the lines relating to SMTPS and Submission near the top (each should have 4-5 lines associated with them).
After that, restart Postfix with "/etc/init.d/postfix restart".
Once you've done that, try configuring your client to connect on port 465 or 587, and see if that does the trick.
-Eric
Well, Eric, you've nailed it agin. Job's a good 'un, as we would say.
The weird thing is, I set up an identical server a couple of weeks ago (same physical server, same OS), and I didn't have to change the mail_location in dovecot.conf - it just wasn't an issue.
Ho hum, rockin' along now so thanks once again.
Must say, I've never found a forum where responses are so forthcoming and also so unerringly accurate. Particularly impressive given the breadth of issues discussed here.
Three months in I'm finding Virtualmin a great product to use, but there's no doubt it would be a scary place without this forum :-) so thanks to all, especially Eric