Postfix: warning: connect to 127.0.0.1:10023: Connection refused, not receiving mail from outside domains

Question

I've had a Postfix server running for a very short amount of time, and it worked, but I had to restart the server today, and I'm no longer receiving e-mails from external sources:

Jan 23 01:34:44 myservername postfix/smtpd[1055]: connect from db3ehsobe006.messaging.microsoft.com[213.199.154.144]
Jan 23 01:34:45 myservername postfix/smtpd[1055]: warning: connect to 127.0.0.1:10023: Connection refused
Jan 23 01:34:45 myservername postfix/smtpd[1055]: warning: problem talking to server 127.0.0.1:10023: Connection refused
Jan 23 01:34:46 myservername postfix/smtpd[1055]: warning: connect to 127.0.0.1:10023: Connection refused
Jan 23 01:34:46 myservername postfix/smtpd[1055]: warning: problem talking to server 127.0.0.1:10023: Connection refused
Jan 23 01:34:46 myservername postfix/smtpd[1055]: NOQUEUE: reject: RCPT from db3ehsobe006.messaging.microsoft.com[213.199.154.144]: 451 4.3.5 Server configuration problem; from=<MyKnownWorking@EmailAccountOutside> to=<[email protected]> proto=ESMTP helo=<db3outboundpool.messaging.microsoft.com>

The server is listening on port 10023, but I noticed it's only listening via IPv6:

> sudo netstat -a | grep 10023
tcp6       0      0 ip6-localhost:10023     [::]:*                  LISTEN

I have no firewall rules in place that would deny that would specific port, hell, I went ahead and flushed the rule set just to confirm it. Here's the output of my postconf -n (I edited out my domain name in place of "mydomain.com":

> sudo postconf -n
alias_database = hash:/etc/aliases
alias_maps = hash:/etc/aliases
    append_dot_mydomain = no
biff = no
broken_sasl_auth_clients = yes
config_directory = /etc/postfix
content_filter = amavis:[127.0.0.1]:10024
disable_vrfy_command = yes
inet_interfaces = all
inet_protocols = ipv4
mailbox_size_limit = 0
message_size_limit = 0
mydestination = localhost.$mydomain, localhost, mail.mydomain.com, servername.mydomain.com
mynetworks = 127.0.0.0/8 [::ffff:127.0.0.0]/104 [::1]/128
mynetworks_style = host
myorigin = /etc/mailname
readme_directory = no
receive_override_options = no_address_mappings
recipient_delimiter = +
relayhost =
smtp_tls_session_cache_database = btree:${data_directory}/smtp_scache
smtp_use_tls = yes
smtpd_banner = mail.mydomain.com ESMTP $mail_name
smtpd_delay_reject = yes
smtpd_helo_required = yes
smtpd_recipient_restrictions = reject_unauth_pipelining, permit_mynetworks, permit_sasl_authenticated, reject_non_fqdn_recipient, reject_unknown_recipient_domain, reject_unauth_destination, check_policy_service inet:127.0.0.1:10023, permit
smtpd_sasl_auth_enable = yes
smtpd_sasl_local_domain =
smtpd_sasl_security_options = noanonymous
smtpd_sender_restrictions = permit_sasl_authenticated, permit_mynetworks, warn_if_reject reject_non_fqdn_sender, reject_unknown_sender_domain, reject_unauth_pipelining, permit
smtpd_tls_cert_file = /etc/ssl/private/mail.mydomain.com.crt
smtpd_tls_key_file = /etc/ssl/private/mail.mydomain.com.key
smtpd_tls_session_cache_database = btree:${data_directory}/smtpd_scache
smtpd_use_tls = yes
virtual_alias_maps = mysql:/etc/postfix/maps/alias.cf
virtual_gid_maps = static:5000
virtual_mailbox_base = /var/spool/mail/virtual
virtual_mailbox_domains = mysql:/etc/postfix/maps/domain.cf
virtual_mailbox_limit = 0
virtual_mailbox_maps = mysql:/etc/postfix/maps/user.cf
virtual_uid_maps = static:5000

As you can see, I'm even trying to specify via inet_protocols that it listens on ipv4 connections. I've tried it with and without that command.

Any help in troubleshooting would be greatly appreciated! And of course, if you see anything in my configuration is glaringly stupid, I am not above advice or criticism.

Answer 1

Your last check on smtpd_recipient_restrictions uses a policy service to verify the recipient. Typically this is a postgrey service and appears to have some problems with Postfix connecting to it.

smtpd_recipient_restrictions = ...,check_policy_service inet:127.0.0.1:10023, permit

If you remove check_policy_service inet:127.0.0.1:10023 from the smtpd_recipient_restrictions, you should eliminate the error, but you still should determine what happen to your postgrey or other service that would be running here.

Checking for Postgrey on an Unbuntu System

Typically a postgrey default config will listen on port 10023 for connections and determine if they should be allowed or rejected. Some pieces on an Unbutu server that you can check for to see if this is installed are...

• Do you have a /etc/default/postgrey file? This is the basic config file.
• Do you have a /etc/postgrey folder? This is where you can whitelist elements.
• When you run > which postgrey does it find a binary? Mine is found in /usr/sbin/postgrey.
• Do you have a /etc/init.d/postgrey script for starting it up on boot up? This is the typical location for Ubuntu daemons.

These will only give you some clues as to if this server may have had postgrey configured at one time. You will need to look further for troubleshooting if the process is not running properly on your server.

Answer 2

Faced with same problem, tried many ways such as proposed by bshea, googling and trying.
Base: Ubuntu 14.04.2 LTS, postgrey service starts, but doesn't appear in process list, i.e. service starts but quietly drops out.

Found solution to change line in /etc/default/postgrey:

Change this line:

POSTGREY_OPTS="--inet=10023"

To this

POSTGREY_OPTS="--inet=127.0.0.1:10023"

No need to play with ports, protocols, neither version-of-something downgrade. Can not explain why, but service is in ps -aux and all works.

Answer 3

Actually, you are not forced to use ipv6, you may set Postgrey and Postfix to ipv4. The problem is, Postgrey (probably version 1.33 and newer) refuse to start on ipv4 localhost ip 127.0.0.1, so you may use your ethX ip address.

In /etc/postfix/main.cf change this:

check_policy_service inet:127.0.0.1:10023

to:

check_policy_service inet:<your_ipv4_address>:10023

Then restart Postfix:

sudo service postfix restart

In /etc/default/postgrey change this:

POSTGREY_OPTS="--inet=10023 --delay=60"

to:

POSTGREY_OPTS="--inet=<your_ipv4_address>:10023 --delay=60"

Then restart Postgrey:

sudo service postgrey restart

Very userful was this Debian mail list message:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=656046#25

Answer 4

The IP:PORT issue isn't/wasn't an issue for me (Ubuntu 14.04.2 LTS). It can be misleading to assume postgrey is breaking/exiting on local port attach/execution failure.

First check to see if daemon is recognized as running :

sudo service postgrey status

If it is not running/recognized, you will still see :

warning: problem talking to server [::1]:10023: Connection refused

in your mail log. Postgrey simply isn't attached/running. If you find it running (may be more than once) and cannot service stop or status it, you probably have a pid/truncate issue. If so, do :

ps a | grep postgrey

Kill (each) process manually. Remove any postgrey.pid files from /var/run (/run).

Assuming all your postgrey files are default and it is trying to start, but failing on reboots/restarts/stops (after checking 'sudo service postgrey status'), you probably have issues with lockfile and named process for stop-start-daemon.

POSTGREY has a VERY basic init.d script which does not do much checking before/after it executes things. The maintainers really need to work on this.

Main problem is start-stop-daemon looking for truncated named file :

( https://bugs.launchpad.net/ubuntu/+source/postgrey/+bug/1289424 )

The script as installed checks for $NAME = "postgrey".
( "--name $NAME" )
Unfortunately, what it needs to use (Ubuntu/Debian) is : '/usr/sbin/postg'
(only 15 chars and is part of path - which doesn't exist obviously.)

The fix (add variable PROCNAME with truncated version including path):

PROCNAME=`echo $DAEMON | cut -c -15`

(Will then check for '/usr/sbin/postg' instead)

and will go under "--name" :

  stop)
        log_daemon_msg "Stopping $DESC" "$NAME"
        if start-stop-daemon --stop --oknodo --quiet \
                --pidfile $PIDFILE --name $PROCNAME

..

  reload|force-reload)
        log_action_begin_msg "Reloading $DESC configuration..."
        if start-stop-daemon --stop --signal 1 --quiet \
                --pidfile $PIDFILE --name $PROCNAME

(Optional)
While I was at it I put the lock file inside it's own directory: '/run/postgrey/postgrey.pid'.

If you prefer it in it's own directory, you need to add this bit to recreate the folder each time:

PIDFOLDER=/var/run/$NAME
PIDFILE=$PIDFOLDER/$NAME.pid

Check for existing dir/folder:

check_pid_dir() {
  if [ ! -d $PIDFOLDER ]; then
    mkdir $PIDFOLDER
    chmod 0755 $PIDFOLDER
  fi
}

and add function to "Start":

case "$1" in
  start)
        check_pid_dir
        log_daemon_msg "Starting $DESC" "$NAME"

My full init.d script is here: https://gist.github.com/bmatthewshea/50e2038563b103e466ea
I have not tested this much - but it is certainly working better than default so far. :)

Answer 5

Davis - you're right, postgrey was changed in 12.04LTS to work in IPv6. Upgrade doesn't change it for postfix and doesn't alert you to make a change.

So you have to do it yourself. Change this:

check_policy_service inet:127.0.0.1:10023

to:

check_policy_service inet:::1:10023

in your /etc/postfix/master.cf file, then restart postfix:

sudo service postfix restart

It will allow postfix to talk IPv6 to postgrey.

Answer 6

This will also break if you've configured Postfix to talk IPv4 only (As I have to due to my ISP still not doing IPv6, and my IPv6 HE tunnel being tagged by services like Netflix, and thus breaking Netflix...)

adding the local host to the INET command flag for postgrey does fix it not binding to IPv4.

I don't get why it can't just bind to BOTH IPv4 AND IPv6 though...