[SOLVED] postgrey / maillog warning: connect to 127.0.0.1:60000: Connection refused

[todd_dsm]

I've installed postgrey on my server per the instructions here.

I sent some test messages: a few from hotmail.com (not whitelisted) and a few from yahoo.com (whitelisted) at approx Thu Oct 29 10:30:00 CDT 2009

Some notes:

# tail -f /var/log/maillog
Oct 29 11:26:21 zerver zmmailboxdmgr[30005]: status requested
Oct 29 11:26:21 zerver zmmailboxdmgr[30005]: status OK
Oct 29 11:27:24 zerver postfix/smtpd[30124]: connect from web65704.mail.ac4.yahoo.com[76.13.9.96]
Oct 29 11:27:24 zerver postfix/smtpd[30124]: warning: connect to 127.0.0.1:60000: Connection refused
Oct 29 11:27:24 zerver postfix/smtpd[30124]: warning: problem talking to server 127.0.0.1:60000: Connection refused
Oct 29 11:27:25 zerver postfix/smtpd[30124]: warning: connect to 127.0.0.1:60000: Connection refused
Oct 29 11:27:25 zerver postfix/smtpd[30124]: warning: problem talking to server 127.0.0.1:60000: Connection refused
Oct 29 11:27:25 zerver postfix/smtpd[30124]: NOQUEUE: reject: RCPT from web65704.mail.ac4.yahoo.com[76.13.9.96]: 451 4.3.5 Server configuration problem; from=<todd50314@yahoo.com> to=<thomas@ptest.us> proto=SMTP helo=<web65704.mail.ac4.yahoo.com>
Oct 29 11:27:25 zerver postfix/smtpd[30124]: disconnect from web65704.mail.ac4.yahoo.com[76.13.9.96]
Oct 29 11:27:49 zerver zmmailboxdmgr[30379]: status requested
Oct 29 11:27:49 zerver zmmailboxdmgr[30379]: status OK

Per the posting in the forum I'm double checked a few things:

1) "One thing I would fix is to get postgrey starting before Zimbra"
Check, I stopped zimbra, restarted postgrey, and started zimbra again: same error shows in logs when hotmail and yahoo attempt to resend.

2) "Have you disabled IPtables ?"
Check, I disabled iptables altogether, then restarted postgrey/zimbra per/ 1): same error shows in logs when hotmail and yahoo attempt to resend.

3) "Found the problem. Couple of .pid files in /opt/zimbra/log were owned by root instead of zimbra".

# ll /opt/zimbra/log/*.pid
-rw-r----- 1 zimbra zimbra 6 Oct 29 11:10 /opt/zimbra/log/convertd.pid
-rw-r----- 1 zimbra zimbra 6 Oct 29 11:10 /opt/zimbra/log/logswatch.pid
-rw-r--r-- 1 zimbra zimbra 5 Oct 29 11:10 /opt/zimbra/log/zmgengraphs.pid
-rw-r--r-- 1 zimbra zimbra 5 Oct 29 11:10 /opt/zimbra/log/zmlogprocess.pid
-rw-r----- 1 zimbra zimbra 5 Oct 29 11:11 /opt/zimbra/log/zmmtaconfig.pid

Check, all pid files in /opt/zimbra/log are owned by zimbra.

Both hotmail and yahoo keep re-attempting but neither have made it through yet. It's been over an hour: Thu Oct 29 11:42:37 CDT 2009

Google isn't providing much in the way of answers. The few postings I did see were to the postgrey mailing list and the answer appears to be that postgrey should either be listening to the unix socket or on 127.0.0.1:60000 but not both. I'm way out of my element here so I may be misunderstanding that piece. Either way:

# netstat -nlp | grep grey
unix  2  [ ACC ]  STREAM  LISTENING  6322171  17767/socket  /var/spool/postfix/postgrey/socket

# grep socket /var/log/messages
Oct 29 11:06:07 zerver postgrey[17767]: Binding to UNIX socket file /var/spool/postfix/postgrey/socket using SOCK_STREAM

# netstat -nlpa | grep 60000
(produces no output)

so that appears to be ok.

I understand this is a postgrey problem and not a zimbra problem but I am submitting this so that there is a zimbra solution for all interested in greylisting.

Thanks in advance,
todd_dsm

Don't forget to Vote for this RFE:
RFE: A place To Display the contents of 'My Documents'
Reasoning: It's new, bold, and cool.

[menno.pieters]

Check your postgrey configuration (or init script). It is obviously not listening to TCP port 60000, but to a UNIX socket.

Make sure postfix starts with an option like:

--inet=127.0.0.1:60000

[todd_dsm]

Well, I'm probably the only one that didn't know this but for posterities sake:

menno.pieters had it tagged right; if you follow the directions from the walk-through it will get you 99% of the way home. The the necessary RHEL5-specific change occurs in the init script for postgrey:

# cat /etc/init.d/postgrey
#!/bin/sh
...
prog=postgrey
postgrey=/usr/sbin/$prog
DBPATH=/var/spool/postfix/postgrey
SOCKET=$DBPATH/socket          # Remove this line
OPTIONS="--unix=$SOCKET"       # Change this line to        OPTIONS="--inet=127.0.0.1:60000 -d"
...

By default, the postgrey rpm that you get at DAG's repo is configured by default to listen on a unix socket. The version I pulled down with yum was postgrey-1.32-1.el5.rf.

After restarting postgrey and running netstat I found this:

# netstat -nlpa | grep post
tcp     0     0     127.0.0.1:60000     0.0.0.0:*     LISTEN     21891/postgrey -d -
unix    2    [ ]	                    DGRAM    7010876     21891/postgrey -d -

The second line makes it appear as if there is still a unix socket being created but there is no LISTEN on that line so it seems to go ignored. I haven't dug through the postgrey perl script yet because I am successfully greylisting!

Thanks menno.pieters!
todd_dsm

Don't forget to Vote for this RFE:
RFE: A place To Display the contents of 'My Documents'
Reasoning: It's new, bold, and cool.