dovecot - Oct 2011 - trouble setting up director, "invalid argument" for connect() call

Hi Timo & everyone,

I'm trying out a 2-node director setup, but I keep getting the following
error:

Oct  3 16:11:29 imapdir1 dovecot: master: Dovecot v2.0.15 starting up
(core dumps disabled)
Oct  3 16:11:34 imapdir1 dovecot: director: Error:
connect(132.198.100.150:9090) failed: Invalid argument
Oct  3 16:11:41 imapdir1 last message repeated 3 times

Both nodes report this error.  The director on each node is listening on
port 9090 just fine:
> [root at imapdir1 ~]# nc imapdir2 9090
> VERSION director        1       0
> ME      127.0.0.1       9090
> [root at imapdir2 ~]# nc imapdir1 9090
> VERSION director        1       0
> ME      127.0.0.1       9090
"doveadm director ring status" always reports failure of the peer:

> [root at imapdir1 ~]# doveadm director ring status
> director ip                                    port type last
> failed        
> 127.0.0.1                                      9090 self
> never              
> 132.198.100.150                                9090      2011-10-04
> 13:28:55
> [root at imapdir2 ~]# doveadm director ring status
> director ip                                    port type last
> failed        
> 127.0.0.1                                      9090 self
> never              
> 132.198.100.149                                9090      2011-10-04
> 13:24:08

Has anyone else seen this?

FWIW, a single-node director works fine, proxying IMAP, LMTP and POP. 

I'm guessing I have a broken config, but if I do I can't figure it out. 
Hope I'm not missing something *too* obvious...

Here's doveconf -n from imapdir1:

 # 2.0.15: /etc/dovecot/dovecot.conf
 # OS: Linux 2.6.18-274.3.1.el5 x86_64 Red Hat Enterprise Linux Server
release 5.7 (Tikanga)
 base_dir = /var/run/dovecot/
 director_mail_servers = penguina.uvm.edu penguinb.uvm.edu penguinc.uvm.edu
 director_servers = imapdir1.uvm.edu imapdir2.uvm.edu
 lmtp_proxy = yes
 login_trusted_networks
132.198.101.211,132.198.101.212,132.198.101.213,132.198.101.214,132.198.101.215,132.198.101.216,132.198.100.45,132.198.100.46,132.198.100.47
 passdb {
   args = proxy=y nopassword=y protocol=smtp
   driver = static
 }
 service auth {
   unix_listener auth-userdb {
     group = mail
     mode = 0660
     user = dovecot
   }
 }
 service director {
   fifo_listener login/proxy-notify {
     mode = 0666
   }
   inet_listener {
     port = 9090
   }
   unix_listener director-userdb {
     mode = 0600
   }
   unix_listener login/director {
     mode = 0666
   }
 }
 service imap-login {
   executable = imap-login director
 }
 service lmtp {
   client_limit = 1
   inet_listener lmtp {
     port = 24
   }
   unix_listener /var/lib/dovecot/lmtp-socket {
     group = root
     mode = 0600
     user = root
   }
 }

Best,
Jim

On 10/4/11 13:31 , Jim Lawson wrote:
> Hi Timo & everyone,
>
> I'm trying out a 2-node director setup, but I keep getting the
following
> error:
>
> Oct  3 16:11:29 imapdir1 dovecot: master: Dovecot v2.0.15 starting up
> (core dumps disabled)
> Oct  3 16:11:34 imapdir1 dovecot: director: Error:
> connect(132.198.100.150:9090) failed: Invalid argument
> Oct  3 16:11:41 imapdir1 last message repeated 3 times
I was specifying DNS names in director_servers (imapdir1.uvm.edu,
imapdir2.uvm.edu).  /etc/hosts listed 127.0.0.1 as the address for the
local host name (in addition to "localhost").  This can happen on some
RHEL installs that begin their life as DHCP clients.

So when the director tried to figure out its self_ip, it always picked
127.0.0.1.  Calling connect() using a localhost socket with a non-local
address as destination just doesn't work so well, I guess.  Seeing
127.0.0.1 as "self" in "doveadm director ring status" should
have been a
clue, but I hadn't seen the output before to have anything to compare it to.

2 easy fixes:  specify real IP addresses in the director_servers config,
or don't list 127.0.0.1 as your local hostname in /etc/hosts.

Jim