Configure a Centos 7 postfix mail server with virtual users

Configure a mail server is one of top sysadmins nightmares, here i will try to make your painful  journey a bit less unpleasant.

Starting from Centos 7 minimal install.

If in any moment something doesn’t work just leave a comment and i will try to give you some feedback

Before run this guide ensure that:

  • Your server has an public ip
  • Your ISP will configure a reverse dns for your public ip
  • You can modify DNS records for all domains that you can attach to your new mailserver

 

Step 1: Enable extra repositories


yum install epel-release

yum -y install http://nginx.org/packages/centos/7/noarch/RPMS/nginx-release-centos-7-0.el7.ngx.noarch.rpm

yum -y update

Step 2: install packages


yum -y install perl-MailTools perl-MIME-EncWords perl-Email-Valid perl-Test-Pod roundcubemail dovecot dovecot-mysql  dovecot-pigeonhole  perl-Mail-Sender perl-Log-Log4perl imapsync offlineimap amavisd-new clamav perl-Razor-Agent mariadb-server opendkim vim wget crypto-utils mod_ssl.x86_64 nginx php php-mysql php-fpm  clamav-update php-imap.x86_64 NetworkManager-tui mailx lrzip lzop lz4 arj  unzoo cabextract p7zip fail2ban php-mcrypt.x86_64


yum install http://pkgs.repoforge.org/unrar/unrar-5.0.3-1.el7.rf.x86_64.rpm

Step 3: Change your servers hostname

To avoid problems with hostname resolution the easy way is use NotworkManager-tui to set you hostname to a FQDN, when you configure your MX registers these point to a hostname, like mail.mydomain.com select your desired FQDN and execute


nmtui

this will show nmtui dialog

nmtui main screenselect Set system hosname

nmtui set hostname dialogput your desired hostname and click OK

finaly exit nmtui

and apply changues


systemctl restart systemd-hostnamed

Step 4: Verify postifx local delivery

At this step we will ensure that postfix can send email to local users firstly we will add a two new users joe and alfred


useradd -d /home/joe -M -N -s /sbin/nologin joe

we will send a local mail


echo "Hello" | mail -s "test" joe@localhost

and verify if mail has been delivered


tail /var/log/maillog

May  6 18:57:34 localhost postfix/pickup[1705]: 3D20F1000ED: uid=0 from=<root>
May  6 18:57:34 localhost postfix/cleanup[2444]: 3D20F1000ED: message-id=<20150506165734.3D20F1000ED@mail.yourfqdn.com>
May  6 18:57:34 localhost postfix/qmgr[1706]: 3D20F1000ED: from=<root@mail.yourfqdn.com>, size=494, nrcpt=1 (queue active)
May  6 18:57:34 localhost postfix/local[2446]: 3D20F1000ED: to=<joe@localhost.yourfqdn.com>, orig_to=<joe@localhost>, relay=local, delay=0.11, delays=0.07/0.03/0/0.01, dsn=2.0.0, status=sent (delivered to mailbox)
May  6 18:57:34 localhost postfix/qmgr[1706]: 3D20F1000ED: removed

if it’s delivered next step is add a domain like centostestmail.com


vim /etc/postfix/main.cf

edit myhostname property near line 77

myhostname= yourfqdndomain

near line  164

mydestination = $myhostname, localhost.$mydomain, localhost, centostestmail.com

finaly execute


systemctl restart postfix.service

verify that postfix is running


systemctl status postfix.service

if postfix is running we will verify if postfix delivers mail to adress with domain centostestmail.com


echo "Hello" | mail -s "test" joe@centostestmail.com

verify if mail has been delivered


tail /var/log/maillog

ay  6 19:58:01 mail postfix/pickup[2818]: 9AC9D10025A: uid=0 from=<root>
May  6 19:58:01 mail postfix/cleanup[2829]: 9AC9D10025A: message-id=<20150506175801.9AC9D10025A@mail.yourfqdn.com>
May  6 19:58:01 mail postfix/qmgr[2819]: 9AC9D10025A: from=<root@mail.yourfqdn.com>, size=459, nrcpt=1 (queue active)
May  6 19:58:01 mail postfix/local[2831]: 9AC9D10025A: to=<joe@centostestmail.com>, relay=local, delay=0.14, delays=0.1/0.04/0/0.01, dsn=2.0.0, status=sent (delivered to mailbox)
May  6 19:58:01 mail postfix/qmgr[2819]: 9AC9D10025A: removed

if your log appears like example we can move to the next step

Step 5: Configure Clam Antivirus

If you want to run a mail server you will be a happy sysadmin if you put an antivirus inspecting incoming mail.  Minimize viruses risk is a plus. In my case I will use clam.


setsebool -P antivirus_use_jit on

we need to configure how clam refreshes his database


vim /etc/sysconfig/freshclam

comment or remove last line

#  FRESHCLAM_DELAY=disabled-warn    # REMOVE ME

now we will make a edit clamav config file


vim /etc/freshclam.conf

remove the line that puts

example

finally update your viruses database


freshclam

after we get an updated antivirus we should check if it’s working as expected for these purposes there’s a string called EICAR that should be detected as a virus always.

download EICAR string


wget  http://www.eicar.org/download/eicar.com

analyze EICAR string


clamscan --infected --remove eicar.com

output should be like these

eicar.com: Eicar-Test-Signature FOUND
eicar.com: Removed.

----------- SCAN SUMMARY -----------
Known viruses: 3802265
Engine version: 0.98.6
Scanned directories: 0
Scanned files: 1
Infected files: 1
Data scanned: 0.00 MB
Data read: 0.00 MB (ratio 0.00:1)
Time: 27.957 sec (0 m 27 s)

Step 6: Configure basic settings in spamassasin

At this point we have an antivirus in our server that is’s updated every 3 hours, viruses is a problem for our users but the BIG problem is spam, most of your users incoming mail will be spam, and our users will complaining about that they can work because lost his precious time removing spam in his incoming mail, another part will be losing her time buying magical blue pills, expensive watches and helping African kings moving her money outside Africa.

Anyway first step is configure selinux


setsebool -P spamassassin_can_network on

enable spamassasin service


systemctl start spamassassin.service
systemctl status spamassassin.service
systemctl enable spamassassin.service

update spamassasin definitions


sa-update

a cron task is added when you install spamassin that will update spamassasin definitions is located at  /etc/cron.d/sa-update

Step 7: Integrate spamassasin and clamav with amavisd

At this point clamav and spamassasin are isolated services for integrate these elements we will use amavisd he will be responsive of deliver mails in order of get a virus and spam check.

Clamav is not a lightweight service instead launch a local copy for every mail, we load a service to be ready faster.

we need to provide some config files


cp /usr/share/doc/clamav-server*/clamd.sysconfig /etc/sysconfig/clamd.amavisd

we need to adapt config file to our actual configuration


vi /etc/sysconfig/clamd.amavisd

CLAMD_CONFIGFILE=/etc/clamd.d/amavisd.conf
CLAMD_SOCKET=/var/run/clamd.amavisd/clamd.sock

we will create a couple of new files


vi /etc/tmpfiles.d/clamd.amavisd.conf

add this content

d /var/run/clamd.amavisd 0755 amavis amavis -

vi /usr/lib/systemd/system/clamd@.service

with this content

[Unit]
Description = clamd scanner (%i) daemon
After = syslog.target nss-lookup.target network.target
[Service]
Type = simple
ExecStart = /usr/sbin/clamd -c /etc/clamd.d/%i.conf --nofork=yes
Restart = on-failure
PrivateTmp = true
[Install]
WantedBy=multi-user.target

now we can enable clamd@amavisd service


systemctl start clamd@amavisd

systemctl enable clamd@amavisd

now we need configure amavisd service


vim /etc/amavisd/amavisd.conf

at line 16 set number of amavisd childrens

more childres uses more ram but delivers more mail at once, one amavisd children consumes near 40% of cpu in a low end server, be careful if you receive a lot of mails at once can be a big punch in your cpu have too many childrens

$max_servers = <a number>

at line 20 set $mydomain we need to put step 3 domain name

$mydomain = ‘desireddomain.com’;

at line 152 aprox set your hostname

$myhostname= ‘mail.desireddomain.com’;


systemctl start amavisd.service

after this we must search in file /var/log/maillog strings like these

Mar 19 17:51:48 mail amavis[21284]: ANTI-SPAM-SA code    loaded

Mar 19 17:51:48 mail amavis[21284]: Using primary internal av scanner code for ClamAV-clamd

to ensure that amavisd detects our antivirus and spamassasin

next step is enable service


systemctl enable amavisd.service

Step 8:Integrate postfix with amavisd

Amavisd will pass all incoming mail to our antivirus and antispam and verify that we are receiving a clean mail, but at this moment we have postfix and amavisd isolated, we need make a small integration.


vim /etc/postfix/main.cf

append this line at the end

content_filter=smtp-amavis:[127.0.0.1]:10024

vim /etc/postfix/master.cf

append these lines at the end

smtp-amavis unix -      -       n       -       6       smtp
        -o smtp_data_done_timeout=1200
        -o smtp_send_xforward_command=yes
        -o disable_dns_lookups=yes
127.0.0.1:10025 inet n  -       n      -        -       smtpd
        -o content_filter=
        -o local_recipient_maps=
        -o relay_recipient_maps=
        -o smtpd_restriction_classes=
        -o smtpd_client_restrictions=
        -o smtpd_helo_restrictions=
        -o smtpd_sender_restrictions=
        -o smtpd_recipient_restrictions=permit_mynetworks,reject
        -o mynetworks=127.0.0.0/8
        -o strict_rfc821_envelopes=yes
        -o smtpd_error_sleep_time=0
        -o smtpd_soft_error_limit=1001
        -o smtpd_hard_error_limit=1000

 

in line smtp-avavis unix – – n – <number> smtp

try that number value be the same that amavisd children

restart postfix service


systemctl stop postfix.service

systemctl start postfix.service

now we need to check that our  postfix and amavisd service are integrated, the easy way is send an spam mail and a virus mail, one with gtube string another with eicar

we need to put EICAR and GTUBE strings in a mail.

EICAR String

Gtube String

test spam and virus mail

after send this couple of test you need to check in /var/log/maillog and output like these

May  7 00:20:14 mail postfix/pickup[4434]: 578BA100322: uid=0 from=&amp;lt;root&amp;gt;
May  7 00:20:14 mail postfix/cleanup[4458]: 578BA100322: message-id=&amp;lt;20150506222014.578BA100322@mail.yourfqdn.com&amp;gt;
May  7 00:20:14 mail postfix/qmgr[4435]: 578BA100322: from=&amp;lt;root@mail.yourfqdn.com&amp;gt;, size=527, nrcpt=1 (queue active)
May  7 00:20:15 mail amavis[2388]: (02388-02) Blocked SPAM {DiscardedOpenRelay,Quarantined}, &amp;lt;root@mail.yourfqdn.com&amp;gt; -&amp;gt; &amp;lt;joe@centostestmail.com&amp;gt;, Message-ID: &amp;lt;20150506222014.578BA100322@mail.yourfqdn.com&amp;gt;, mail_id: TcgB4ovK5t1h, Hits: 999.999, size: 527, 1596 ms
May  7 00:20:15 mail postfix/smtp[4460]: 578BA100322: to=&amp;lt;joe@centostestmail.com&amp;gt;, relay=127.0.0.1[127.0.0.1]:10024, delay=1.6, delays=0.03/0/0.01/1.6, dsn=2.7.0, status=sent (250 2.7.0 Ok, discarded, id=02388-02 - spam)
May  7 00:20:15 mail postfix/qmgr[4435]: 578BA100322: removed
May  7 00:20:22 mail postfix/pickup[4434]: 4D335100323: uid=0 from=&amp;lt;root&amp;gt;
May  7 00:20:22 mail postfix/cleanup[4458]: 4D335100323: message-id=&amp;lt;20150506222022.4D335100323@mail.yourfqdn.com&amp;gt;
May  7 00:20:22 mail postfix/qmgr[4435]: 4D335100323: from=&amp;lt;root@mail.yourfqdn.com&amp;gt;, size=527, nrcpt=1 (queue active)
May  7 00:20:22 mail clamd[1084]: /var/spool/amavisd/tmp/amavis-20150507T001929-02391-G5h5_dlo/parts/p002: Eicar-Test-Signature FOUND
May  7 00:20:22 mail clamd[1084]: /var/spool/amavisd/tmp/amavis-20150507T001929-02391-G5h5_dlo/parts/p001: Eicar-Test-Signature FOUND
May  7 00:20:22 mail amavis[2391]: (02391-02) Blocked INFECTED (Eicar-Test-Signature) {DiscardedOpenRelay,Quarantined}, &amp;lt;root@mail.yourfqdn.com&amp;gt; -&amp;gt; &amp;lt;joe@centostestmail.com&amp;gt;, Message-ID: &amp;lt;20150506222022.4D335100323@mail.yourfqdn.com&amp;gt;, mail_id: yIy3drLObXjq, Hits: -, size: 527, 134 ms
May  7 00:20:22 mail postfix/smtp[4460]: 4D335100323: to=&amp;lt;joe@centostestmail.com&amp;gt;, relay=127.0.0.1[127.0.0.1]:10024, delay=0.16, delays=0.02/0/0.01/0.13, dsn=2.7.0, status=sent (250 2.7.0 Ok, discarded, id=02391-02 - INFECTED: Eicar-Test-Signature)
May  7 00:20:22 mail postfix/qmgr[4435]: 4D335100323: removed

Step 9: Configure mariadb

At this point we have a working postfix mail server with antivirus and antispam, at this point we can add local user accounts and they will get a mail account. But i prefer to have all users in a database and foget to have hundreds of accounts


vim /etc/my.cnf.d/server.cnf

#
# These groups are read by MariaDB server.
# Use it for options that only the server (but not clients) should see
#
# See the examples of server my.cnf files in /usr/share/mysql/
#

# this is read by the standalone daemon and embedded servers
[server]
innodb_file_per_table
innodb_file_format = Barracuda
# this is only for the mysqld standalone daemon
[mysqld]

# this is only for embedded server
[embedded]

# This group is only read by MariaDB-5.5 servers.
# If you use the same .cnf file for MariaDB of different versions,
# use this group for options that older servers don't understand
[mysqld-5.5]

# These two groups are only read by MariaDB servers, not by MySQL.
# If you use the same .cnf file for MySQL and MariaDB,
# you can put MariaDB-only options here
[mariadb]

[mariadb-5.5]

enable mariadb service


systemctl  enable mariadb.service

Start mariadb database server


systemctl  start mariadb.service

secure mariadb installation


mysql_secure_installation

follow screen instructions

Optional: if you will need access to database open port using this command


firewall-cmd --permanent --add-service=mysql
firewall-cmd --reload

Remember configure a backup script for your database

Step 10: Generate SSL Certificate


genkey --days 3650 mail.yourdamain.com

follow on screen instructions and this will generate two files.

private part of certificate at

/etc/pki/tls/certs/mail.yourdomain.com.0.csr

and public part of certificate at

/etc/pki/tls/private/mail.yourdomain.com.key

you should get your certificate signed by a certification autority search for one pay and follor certification authority instructions 🙁

Step 11: Create a user account and a folder were store  mails


mkdir /var/vmail

chmod 770 /var/vmail/

useradd -r -u 101 -g mail -d /var/vmail/ -s /sbin/nologin -c &amp;quot;Virtual Mailbox&amp;quot; vmail

chown vmail:mail /var/vmail/

Step 12: Configure nginx

Fist step is verify that nginx server is installed and working


systemctl enable nginx.service
systemctl start nginx.service

firewall-cmd --permanent --zone=public --add-service=http
firewall-cmd --permanent --zone=public --add-service=https
firewall-cmd --reload

select your favorite webbrowser and navigate to your server ip

you should see something like this

welcome nginx

Step 13: Configure php


vim  /etc/php.ini

line 763 should be

cgi.fix_pathinfo=0

line 878 should be your timezone

search here http://php.net/manual/en/timezones.php

date.timezone = Continent/Country

Step 14: Configure PostfixAdmin

in my case i will create a domain like mailadmin.yourdomain.com

we have to download postfixadmin


wget http://downloads.sourceforge.net/project/postfixadmin/postfixadmin/postfixadmin-2.92/postfixadmin-2.92.tar.gz

and move to a valid ubication


mv postfixadmin-2.92.tar.gz /var/www

cd /var/www

tar xzvf postfixadmin-2.92.tar.gz

rm postfixadmin-2.92.tar.gz

mv postfixadmin-2.92/ postfixadmin.yourdomain.com

chown -R  nginx:nginx postfixadmin.yourdomain.com

mkdir /var/lib/php/postfixadmin.yourdomain.com

disable php-fpm default socket


mv  /etc/php-fpm.d/www.conf /etc/php-fpm.d/www.conf.dis

create a php-fpm postfixadmin socket


vim /etc/php-fpm.d/postfixadmin.yourdomain.com.conf

[postfixadmin.yourdomain.com]
listen = /var/run/php-fpm/postfixadmin.yourdomain.com.socket
listen.allowed_clients = 127.0.0.1
user = nginx
group = nginx
pm = dynamic
pm.max_children = 50
pm.start_servers = 5
pm.min_spare_servers = 3
pm.max_spare_servers = 35
slowlog = /var/log/php-fpm/postfixadmin.log
rlimit_files = 1024
rlimit_core = 0
chdir = /var/www/postfixadmin.yourdomain.com
php_flag[display_errors] = off
php_admin_value[error_log] = /var/log/php-fpm/postfixadmin.yourdomain.com-error.log
php_admin_flag[log_errors] = on
php_admin_value[memory_limit] = 128M
php_value[session.save_handler] = files
php_value[session.save_path] = /var/lib/php/postfixadmin.yourdomain.com

start php-fpm service


systemctl start  php-fpm.service

check if php-fpm has started


systemctl status  php-fpm.service

output should appear like these

php-fpm.service - The PHP FastCGI Process Manager
   Loaded: loaded (/usr/lib/systemd/system/php-fpm.service; disabled)
   Active: active (running) since Thu 2015-05-07 00:52:05 CEST; 2s ago
 Main PID: 5499 (php-fpm)
   Status: ;Ready to handle connections;
   CGroup: /system.slice/php-fpm.service
           ├─5499 php-fpm: master process (/etc/php-fpm.conf)
           ├─5501 php-fpm: pool postfixadmin.yourdomain.com
           ├─5502 php-fpm: pool postfixadmin.yourdomain.com
           ├─5503 php-fpm: pool postfixadmin.yourdomain.com
           ├─5504 php-fpm: pool postfixadmin.yourdomain.com
           └─5505 php-fpm: pool postfixadmin.yourdomain.com

May 07 00:52:05 yourfqdn systemd[1]: Started The PHP FastCGI Process Manager.

enable php-fpm service to start at boot


systemctl enable  php-fpm.service

Next we need to create postfix user database from a shell execute:


mysql -u root -p -e &amp;quot;CREATE DATABASE postfix;&amp;quot;

mysql -u root -p -e &amp;quot;CREATE USER postfix@localhost IDENTIFIED BY 'put_here_your_password';&amp;quot;

mysql -u root -p -e &amp;quot;GRANT ALL PRIVILEGES ON postfix . * TO postfix@localhost;&amp;quot;

at this point we need to create a host for postfixadmin in nginx

I will use a selfsigned certificate for https you can pay a certification authority to bypass bad certificate errors in your browser


genkey --days 3650 postfixadmin.yourdomain.com

follow screen instructions, if you put a password in your keys every time that you reboot your system you will need to put that

trick- collecting random data could take a lot of time, to acelerate this process you can dowload a linux distro iso to acelerate this process, it reads data from network card


vim /etc/nginx/conf.d/put_here_your_postfixadmin_domain.conf

 

server {

  listen 80;
  server_name postfixadmin.yourfqdn.com;
  error_log  /var/log/nginx/postfixadmin.yourfqdn.com.error.log warn;
  access_log  /var/log/nginx/postfixadmin.yourfqdn.com.access.log;

  return 301 https://$server_name$request_uri; # enforce https

}

server {

   listen          443 ssl;
   server_name     postfixadmin.yourfqdn.com;

 error_log /var/log/nginx/postfixadmin.yourfqdn.com.secure.error.log warn;
 access_log /var/log/nginx/postfixadmin.yourfqdn.com.secure.access.log;
   root            /var/www/postfixadmin.yourfqdn.com;
   index           index.php;
   charset         utf-8;

   ## SSL settings
   ssl_certificate           //etc/pki/tls/certs/postfixadmin.yourfqdn.com.crt;
   ssl_certificate_key       /etc/pki/tls/private/postfixadmin.yourfqdn.com.key;
   ssl_protocols             TLSv1.2;
   ssl_ciphers               &amp;quot;EECDH+ECDSA+AESGCM:EECDH+aRSA+AESGCM:!aNULL:!eNULL:!LOW:!3DES:!MD5:!EXP:!PSK:!SRP:!DSS:!RC4&amp;quot;;
   ssl_prefer_server_ciphers on;
   ssl_session_cache         shared:SSL:10m;
   ssl_session_timeout       10m;
   ssl_ecdh_curve            secp521r1;

   add_header Strict-Transport-Security max-age=31536000;
   # add_header X-Frame-Options DENY;

   # auth_basic &amp;quot;Restricted area&amp;quot;;
   # auth_basic_user_file /etc/nginx/passwd;

   location / {
      try_files $uri $uri/ index.php;
   }

   location ~* \.php$ {
        try_files $uri =404;
        fastcgi_split_path_info ^(.+\.php)(/.+)$;
        include       fastcgi_params;
        fastcgi_pass  unix:/var/run/php-fpm/postfixadmin.yourfqdn.com.socket;
        fastcgi_index index.php;
        fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
   }

}

 

we need to allow in selinux communication between nginx and php-fpm


setsebool -P httpd_can_network_connect on

and allow selinux that permits nginx to write files in server


setsebool -P httpd_unified on

and allow php-fpm to write session files


chown nginx:nginx /var/lib/php/postfixadmin.yourdomain.com

now configure postfixadmin parameters


vim /var/www/postfixadmin.yourdomain.com/config.inc.php

it’s self explanatory but you will need to changue these parameters


$CONF['configured'] = true;

$CONF['setup_password'] = 'puthereastrongpassword';

$CONF['database_type'] = 'mysqli';
$CONF['database_host'] = 'localhost';
$CONF['database_user'] = 'postfix';
$CONF['database_password'] = 'yourpassword';
$CONF['database_name'] = 'postfix';

$CONF['emailcheck_resolve_domain']='NO';

$CONF['default_aliases'] = array (
    'abuse' =&amp;gt; 'abuse@amaindomainthatyouwilluse.com',
    'hostmaster' =&amp;gt; 'hostmaster@amaindomainthatyouwilluse.com',
    'postmaster' =&amp;gt; 'postmaster@amaindomainthatyouwilluse.com',
    'webmaster' =&amp;gt; 'webmaster@amaindomainthatyouwilluse.com'
);

we will need restart nginx again to enable postfixadmin site


systemctl restart nginx.service

now we can setup our postfixadmin instance

open in your browser

https://postfixadmin.yourdomain.com/setup.php

postfixadmin setupfill all data and follow instructions

finally access to  https://postfixadmin.yourdomain.com and login in order to configure your domains

postfixadmin main screen

we will create one test domain like newcentostestmail.com

in tab Domain List we will select new domain

postixadmin add domainafter add new domain we will create a mailbox

go to Domain List -> Domain List

click on newcentostestmail.com

an Add Mailbox and create joe@newcentostestmail.com mailbox

postfixadmin create mailboxfinally we have created a mailbox we need to link our database with postfix

step 15: Cleanup some test config

Do you want to have a user in your system called joe? you don’t


userdel joe

Do you want map mails from @centostestmail.com? neither

remove centostestmail.com from mydestination values


vim /etc/postfix/main.cf

at line 164 approx leave these values for mydestination

mydestination = $myhostname, localhost.$mydomain, localhost

restart postfix


systemctl restart postfix

step 16: Connect postfix with mariadb:

we have a database that contains a domain list and and a list of emails we need that postfix can access to database in order to read all data

because postfix and other components only will need read permisions we will create a only read user


mysql -u root -p -e &amp;quot;CREATE USER postfixread@localhost IDENTIFIED BY 'put_here_your_password';&amp;quot;

mysql -u root -p -e &amp;quot;GRANT SELECT ON postfix . * TO postfixread@localhost ;&amp;quot;

we will need to link three elements domains, mailboxes and aliases

we will start with domains


vim /etc/postfix/mysql-virtual_domains_maps.cf

using this content

hosts = localhost
user = postfixread
password = put_here_your_password
dbname = postfix
query = SELECT domain FROM domain WHERE domain='%s' AND backupmx = '0' AND active = '1'

vim /etc/postfix/mysql-relay_domains_maps.cf

using this content

hosts = localhost
user = postfixread
password = put_here_your_password
dbname = postfix
query = SELECT domain FROM domain WHERE domain='%s' and backupmx = '1'

next step are mailboxes


vim /etc/postfix/mysql-virtual_mailbox_maps.cf

with this content

hosts = localhost
user = postfixread
password = put_here_your_password
dbname = postfix
query = SELECT maildir FROM mailbox WHERE username='%s' AND active = '1'

vim /etc/postfix/mysql-virtual_mailbox_limit_maps.cf

hosts = localhost
user = postfixread
password = put_here_your_password
dbname = postfix
query = SELECT quota FROM mailbox WHERE username='%s' and active=1

and finally we need to configure aliases


vim /etc/postfix/mysql-virtual_alias_maps.cf

hosts = localhost
user = postfixread
password = put_here_your_password
dbname = postfix
query = SELECT goto FROM alias WHERE address='%s' AND active = '1'

now we need to link our sql queries with postfix config file


vim /etc/postfix/main.cf

at line 297 we need to paste this

relay_domains = proxy:mysql:/etc/postfix/mysql-relay_domains_maps.cf

at line 399 we need to paste this,

virtual_alias_maps = proxy:mysql:/etc/postfix/mysql-virtual_alias_maps.cf,
                     regexp:/etc/postfix/virtual_regexp
virtual_mailbox_base = /var/vmail
virtual_mailbox_domains = proxy:mysql:/etc/postfix/mysql-virtual_domains_maps.cf
virtual_mailbox_maps = proxy:mysql:/etc/postfix/mysql-virtual_mailbox_maps.cf
virtual_mailbox_limit_maps = proxy:mysql:/etc/postfix/mysql-virtual_mailbox_limit_maps.cf
virtual_minimum_uid = 101
virtual_uid_maps = static:101
virtual_gid_maps = static:12

proxy_read_maps = $local_recipient_maps $mydestination $virtual_alias_maps $virtual_alias_domains $virtual_mailbox_maps $virtual_mailbox_domains $relay_recipient_maps $relay_domains $canonical_maps $sender_canonical_maps $recipient_canonical_maps $relocated_maps $transport_maps $mynetworks $smtpd_sender_login_maps $sender_bcc_maps $recipient_bcc_maps $smtp_generic_maps $lmtp_generic_maps $alias_maps $virtual_mailbox_limit_maps

finally we need to create virtual_regexp file


touch /etc/postfix/virtual_regexp

if we restart postfix no problems should appear


systemctl restart postfix.service

if we check our maillog file


tail -f /var/log/maillog

we should see something like this

May  9 17:07:56 mail postfix/postfix-script[2967]: starting the Postfix mail system
May  9 17:07:56 mail postfix/master[2969]: daemon started -- version 2.10.1, configuration /etc/postfix

postfix has started and aparently it has been linked to our database but it’s better verify this

for this we will use postfixadmin but before we need to allow write in our virtual mail storage folder because our directory is not labeled


chcon -R -t mail_spool_t /var/vmail

now we will send a test email to a created mailbox in postfixadmin


echo &amp;quot;Hello&amp;quot; | mail -s &amp;quot;test&amp;quot; joe@newcentostestmail.com

in your maillog you should view something like these


tail /var/log/maillog

May  9 17:10:03 mail postfix/pickup[2970]: 1FD7C10041D: uid=0 from=&amp;lt;root&amp;gt;
May  9 17:10:03 mail postfix/cleanup[2992]: 1FD7C10041D: message-id=&amp;lt;20150509151003.1FD7C10041D@mail.yourfqdn.com&amp;gt;
May  9 17:10:03 mail postfix/qmgr[2971]: 1FD7C10041D: from=&amp;lt;root@mail.yourfqdn.com&amp;gt;, size=462, nrcpt=1 (queue active)
May  9 17:10:04 mail postfix/smtpd[3000]: connect from localhost[127.0.0.1]
May  9 17:10:04 mail postfix/smtpd[3000]: 295DD1004F0: client=localhost[127.0.0.1]
May  9 17:10:04 mail postfix/cleanup[2992]: 295DD1004F0: message-id=&amp;lt;20150509151003.1FD7C10041D@mail.yourfqdn.com&amp;gt;
May  9 17:10:04 mail postfix/qmgr[2971]: 295DD1004F0: from=&amp;lt;root@mail.yourfqdn.com&amp;gt;, size=943, nrcpt=1 (queue active)
May  9 17:10:04 mail amavis[2722]: (02722-01) Passed CLEAN {RelayedOpenRelay}, &amp;lt;root@mail.yourfqdn.com&amp;gt; -&amp;gt; &amp;lt;joe@newcentostestmail.com&amp;gt;, Message-ID: &amp;lt;20150509151003.1FD7C10041D@mail.yourfqdn.com&amp;gt;, mail_id: Yppl42y970Wc, Hits: -0.001, size: 462, queued_as: 295DD1004F0, 975 ms
May  9 17:10:04 mail postfix/smtp[2997]: 1FD7C10041D: to=&amp;lt;joe@newcentostestmail.com&amp;gt;, relay=127.0.0.1[127.0.0.1]:10024, delay=1.2, delays=0.19/0.04/0.04/0.96, dsn=2.0.0, status=sent (250 2.0.0 from MTA(smtp:[127.0.0.1]:10025): 250 2.0.0 Ok: queued as 295DD1004F0)
May  9 17:10:04 mail postfix/qmgr[2971]: 1FD7C10041D: removed
May  9 17:10:04 mail postfix/virtual[3003]: 295DD1004F0: to=&amp;lt;joe@newcentostestmail.com&amp;gt;, relay=virtual, delay=0.09, delays=0.02/0.04/0/0.03, dsn=2.0.0, status=sent (delivered to maildir)
May  9 17:10:04 mail postfix/qmgr[2971]: 295DD1004F0: removed

Hurrah we have virtual mail

your sent mail should be at /var/vmail/newcentostestmail.com/joe/new

step 17: open smtp port, 25, and enable remote access in postfix

Our mailserver accepts connections from port 25 but only in localhost interface, that means that we can’t receive mail from a outside server and our users can’t send email from theirs applications.

firs of all we need to tell to postfix that listens from all interfaces


vim /etc/postfix/main.cf

at line 116 replace content with this

inet_interfaces = all

and restart postfix


systemctl restart postfix.service

postfix is listening at port 25 but firewalld is blocking any attemp from outside, we need to open port 25 in firewalld we just need to execute these commands


firewall-cmd --permanent --zone=public --add-service=smtp

firewall-cmd --reload

at this moment postfix should be contacted from any computer in the internet.

users, spammers, phishers and west Europe sexy women are ready to send tons of emails, we should check that this is correct.

from another computer execute this command


telnet &amp;lt;mailserver ip&amp;gt; 25

and you should get this output

Trying mailserver_ip...
Connected to mailserver_ip.
Escape character is '^]'.
220 mail.yourfqdn.com ESMTP Postfix

Eureka we can receive mails from a lot of people.

step 18: enable smtp security

At this moment if   user try send an email using your new server, it only has support for smtp that is a plain protocol, that means that is really easy read communication contents and get his password. we don’t want this for us.

A long time ago, at the beginning of secure smtp times, that’s 1997. Some people decided that best way to  secure smtp will be wrap it over a ssl or tls connection, that means that all connections should be realized starting a ssl/tls session since first data packet.

To differentiate secure and insecure smtp servers the best way it use a port for insecure smtp,  25, and another for secure connections 465. This strategy is a bit irritating because if another server want deliver a email in a secure mode it should verify if port 465 it’s opened.

Two years after some people  decided that the best way  should be implement STARTTLS this allows to  secure our communication  over a plain  textcommunication at  port 25.

using port 465 smtmps is deprecated and you shouldn’t give support.

if you remember at step 6 you generated a certificate located at

/etc/pki/tls/private/mail.yourdomain.com.key
/etc/pki/tls/certs/mail.yourdomain.com.0.csr

it’s recommendable have our certificate signed by and certification authority,  sorry you must pay 🙁  . Otherwise you can generate a self-signed certificate which will you a lot of security warnings in mail clients.

openssl x509 -req -days 365 -in /etc/pki/tls/certs/mail.yourdomain.com.0.csr -signkey /etc/pki/tls/private/mail.yourdomain.com.key -out /etc/pki/tls/certs/mail.yourdomain.com.crt

we need to tell to postfix where are these certificates and enable smtps


vim /etc/postfix/main.cf

[text]

at the end of the file we need to append this content at the end of file

smtpd_tls_key_file = /etc/pki/tls/private/mail.yourfqdn.com.key
smtpd_tls_cert_file = /etc/pki/tls/certs/mail_yourfqdn_com.crt
smtpd_tls_CAfile = /etc/pki/tls/certs/intermediate.mail.yourfqdn.com.cer

postfix also needs to know that we want to use starttls appending this line

smtpd_tls_security_level = may

Once when we have our certificate variables setted we need restart our postfix service


systemctl restart postfix.service

and check /var/log/maillog output


tail  /var/log/maillog

May  9 17:22:12 mail postfix/postfix-script[3614]: starting the Postfix mail system
May  9 17:22:12 mail postfix/master[3616]: daemon started -- version 2.10.1, configuration /etc/postfix

 

now we need to verify that our smtp port supports STARTTLS this can be done from another computer using telnet.


telnet mailserverip 25

Trying ...
Connected to .
Escape character is '^]'.
220 mail.yourfqdn.com ESMTP Postfix
ehlo testing
250-mail.mail.yourfqdn.com
250-PIPELINING
250-SIZE 10240000
250-VRFY
250-ETRN
250-STARTTLS
250-ENHANCEDSTATUSCODES
250-8BITMIME
250 DSN

You should have a line that puts 250-STARTTLS startttls is enabled

now we will check if ssl works


openssl s_client -starttls smtp -crlf -connect your_server_ip:25

this command should give info about your ssl certificate

step 19: Secure smtp ssl

Unfortunately during last months openssl and his protocols appears to be broken, this security problems should be resolved using two ways:

  • maintain your Openssl packages updated.
  • disable broken protocols and cyphers.

if you’re following this guide I assume that you know  how to use yum update, next step is about cyphers and broken protocols.

openssl supports these protocols:

  • SSLv2 broken
  • SSLv3 broken https://poodle.io/
  • TLS 1.0 Use for interoperability purposes where needed Has known issues that cannot be mitigated
  • TLS 1.1 Use for interoperability purposes where needed  Does not support modern cipher suites.
  • TLS 1.2 Recommended version. Supports the modern AEAD cipher suites.

These protocols use different ciphers

  • Minimal encryption should be 128bits.
  • ADH  Anonymous Diffie-Hellman doesn’t provide authentication
  • NULL no encryption no party
  • Export key exchange suites can be broken easily
  • RC4 is prohibited RFC7465
  • 3DES uses 112bits it’s strong but have performance problems avoid it

At this moment your STARTTLS configuration allow insecure protocols and insecure ciphers we need to fix this.

Simply append these content in your main.cf file

#Disable sslv2 ad SSLv3
smtpd_tls_protocols= !SSLv2, !SSLv3
smtpd_tls_mandatory_protocols= !SSLv2, !SSLv3

#set minimum TLS ciphers grade for tls
smtpd_tls_mandatory_ciphers = high

#use server ciphers instead client preference
tls_preempt_cipherlist = yes

#ciphers to exclude
smtpd_tls_mandatory_exclude_ciphers = aNULL, MD5 , DES, ADH, RC4, PSD, SRP, 3DES, eNULL

with this configuration you just avoided SSLv2, SSLv3 and remove insecure ciphers from TLS1.1

You should be tuned about security problems with openSSL.

step 20: enable Perfect forward secrecy  in postfix

We have secured our SSL/TLS configuration? Yes

we can do more for our security? Yes

HOW? enabling Perfect forward secrecy.

perfect forward secrecy generates a  public  keys per session that means theoretically if our private key is compromised your information is secret.

There are two options

Prime-field groups (EDH) – Server uses a large prime number and a generator

Elliptic-curve groups (EECDH) – Instead a large prime serve uses a elliptic curve algorithm

To configure EDH we need to create  some keys


openssl gendh -out /etc/postfix/dh_512.pem -2 512

openssl gendh -out /etc/postfix/dh_1024.pem -2 1024

openssl gendh -out /etc/postfix/dh_2048.pem -2 2048

append this lines in your /etc/postfix/main.cf


vim /etc/postfix/main.cf

#Perfect forwatd secrecy Prime Field Groups EDH
smtpd_tls_dh1024_param_file = ${config_directory}/dh_2048.pem
smtpd_tls_dh512_param_file = ${config_directory}/dh_512.pem

now we can restart postfix service


systemctl reload postfix.service

for security reasons we should generate a new group of prime numbers daily or hourly, we will create a bash script in order to generate these prime numbers every day


vim /etc/cron.daily/postfix_pfs_edh_regenerate

and copy this content

#!/bin/bash
cd /etc/postfix
umask 022
for legth in 512 1024 2048
do
openssl dhparam -out dh_$legth.tmp $legth &amp;amp;&amp;amp; mv dh_$legth.tmp dh_$legth.pem
chmod 644 dh_$legth.pem
done

and give correct permissions

chmod 700 /etc/cron.daily/postfix_pfs_edh_regenerate

To configure EECDH we should add these lines in main.cf

vim /etc/postfix/main.cf

#Perfect forward secrecy Elliptic-curve Groups EECDH
smtpd_tls_eecdh_grade = strong
tls_eecdh_strong_curve = prime256v1
tls_eecdh_ultra_curve = secp384r1

now we can restart postfix service

systemctl restart postfix.service

Step 21: enable POP and IMAP services

Our users need to read their incoming mail in laptops, mobile devices, fridges….

with smtp user can send mails but can’t read new emails and maintain philosophical conversations about last season most viewed soap opera.

Receiving emails from server can be done using POP, IMAP or both

In my case i preffer use only IMAP but i will both configurations

POP is older than IMAP if your users reads email from multiple devices IMAP should be better, check users email volume and number of devices to select what configuration fits better for each one, new soap operas are really intense.

For provide POP and IMAP connectivity we will use dovecot, Dovecot also includes SASL support for centralized logins, Tradicionally tutorials uses cyrus-sasl to implement authentication, but dovecot includes a SASL implementation, i will try to use it instead include cyrus-sasl package.

Dovecot configuration files are located at /etc/dovecot/conf.d

 Step  22: create config files to Link dovecot with MariaDB user database

With postfixadmin we create a mariadb database, we should give access to dovecot in order to get these data

we need a file with needed queries to provide data about users

vim /etc/dovecot/conf.d/dovecot-mysql.conf.ext

driver = mysql
connect = host=localhost dbname=postfix user=postfixread password=yourpassword

password_query = SELECT username as user, password, concat('/var/vmail/', maildir) as userdb_home,  concat('maildir:/var/vmail/', maildir) as userdb_mail, 101 as userdb_uid, 12 as userdb_gid FROM mailbox  WHERE username = '%u' AND active = '1'

user_query = SELECT concat('/var/vmail/', maildir) as home, concat('maildir:/var/vmail/', maildir) as mail,  101 AS uid, 12 AS gid, CONCAT('*:bytes=', quota) as quota_rule FROM mailbox WHERE  username = '%u' AND active = '1'

also we need a file to provide information about accounts quota

vim /etc/dovecot/conf.d/dovecot-mysql-quota.conf.ext

connect = host=localhost dbname=postfix user=postfixread password=your_password
map {
  pattern = priv/quota/storage
  table = quota2
  username_field = username
  value_field = bytes
}
map {
  pattern = priv/quota/messages
  table = quota2
  username_field = username
  value_field = messages
}

with these files dovecot will query  mariadb info about users and mailbox quotas.

step 23: configure dovecot ssl protocols

now we will configure some setting in dovecot main config file

vim /etc/dovecot/dovecot.conf

line 24 tells dovecot what protocols  should serve it needs to be like

protocols = imap pop3

line 30 is what interfaces where dovecot will be listening

listen = *, ::

line 67 defines behavior when reboot dovecot service

shutdown_clients = yes

now we will restart dovecot service

 systemctl restart dovecot.service

and verify if service has started correctly

tail /var/log/maillog

you should see a line like these

May 10 15:42:02 mail dovecot: master: Dovecot v2.2.10 starting up for imap (core dumps disabled)

 step 24: Configure Dovecot users authentification

Now dovecot is listening in desired ports, but blocked by firewalld, before open these ports we need to do some modifications

When a user request their emails using pop3 or imap, he will need to pass his username and password, there are several ways to send login information:

plain: passed as clear text, secure if is send over a TLS/SSL conection

login: another clear text mechanism, used by outlook clients, secure if is send over a TLS/SSL conection

CRAM-MD5: not clear text, uses HMAC-MD5

these are main supported  authentication protocols

by default dovecot only uses PLAIN mechanism, to add more mechanism we need to edit /etc/dovecot/conf.d/10-auth.conf file

vim /etc/dovecot/conf.d/10-auth.conf

at line 100 the content should be these:

auth_mechanisms = plain login cram-md5

now dovecot should support all these auth_mechanisms but we need to restart our dovecot service to check that

 systemctl restart dovecot.service

and verify if service has restarted correctly

tail /var/log/maillog

you should see a line like these

May 10 15:44:57 mail dovecot: master: Warning: Killed with signal 15 (by pid=1 uid=0 code=kill)
May 10 15:44:57 mail dovecot: master: Dovecot v2.2.10 starting up for imap (core dumps disabled)

step 25: configure Dovecot SSL

like postfix dovecot needs to configure SSL. Dovecot supports these modes:

  • POP3: insecure port 110
  • POP3 with StartTLS: secure port 110
  • POP3S:  POP3 wraped over a TLS/SSL connection 995
  • IMAP: insecure port 143
  • IMAP  with StartTLS: secure port 143
  • IMAPS: IMAP wraped over a TLS/SSL connection 993

some people disable port 110 and port 143 because some email clients make a plain login even if communication over port 143 indicates that only use of    StartTLS is avaliable.

What you should do with 110 and 143 is your decision, it depends of what kind of email clients you support. In mi case I will configure POP3 with StartTLS, POP3S, IMAP with StartTLS and IMAPS

We need to provide info about certificates to dovecot

vim /etc/dovecot/conf.d/10-ssl.conf

and replace certificates configuration chains

ssl_key = &amp;lt;/etc/pki/tls/private/mail.yourdomain.com.key
ssl_cert = &amp;lt;/etc/pki/tls/certs/mail_yourdomaiun_com.crt
ssl_ca = &amp;lt;/etc/pki/tls/certs/intermediate.mail.yourdomain.com.cer

as you learn configuring postfix certificates providing only certificates allow using insecure protocols, we should hardening dovecot configuration near line 52

vim /etc/dovecot/conf.d/10-ssl.conf

# SSL protocols to use
ssl_protocols = !SSLv2 !SSLv3
# SSL ciphers to use
ssl_cipher_list = EECDH+ECDSA+AESGCM:EECDH+aRSA+AESGCM:EECDH+ECDSA+SHA384:EECDH+ECDSA+SHA256:EECDH+aRSA+SHA384:EECDH+aRSA+SHA256:EECDH+aRSA+RC4:EECDH:EDH+aRSA:!aNULL:!eNULL:!LOW:!3DES:!MD5:!EXP:!PSK:!SRP:!DSS:!RC4

# Prefer the server's order of ciphers over client's.
ssl_prefer_server_ciphers = yes

step 26: configure Dovecot mailbox location

Now dovecot uses secure ssl ciphers but need to know were are located accounts mailboxes, this configuration options are located in:

/etc/dovecot/conf.d/10-mail.conf

edit this file

vim /etc/dovecot/conf.d/10-mail.conf

and configure these parameters

mail_location = maildir:/var/vmail/%d/%n/:INDEX=/var/vmail/%d/%n/indexes
mail_uid =101
mail_gid =12
first_valid_uid = 101
last_valid_uid = 101
first_valid_gid = 12
last_valid_gid = 12

now dovecot knows where are located mailboxes

step 27: configure Dovecot pop3 pop3s imap and imaps

Now dovecot needs to know what protocols will  serve and how

vim /etc/dovecot/conf.d/10-master.conf

 

service imap-login {
  inet_listener imap {
    port = 143
  }
  inet_listener imaps {
    port = 993
    ssl = yes
  }

service pop3-login {
  inet_listener pop3 {
    port = 110
  }
  inet_listener pop3s {
    port = 995
    ssl = yes
  }
}

service auth {

  unix_listener /var/spool/postfix/private/auth {
    mode = 0666
    user = vmail
    group = mail
  }

we need to advertise to dovecot that we will only accept login over encrypted connections

vim /etc/dovecot/conf.d/10-auth.conf

disable_plaintext_auth = yes

and new need to link dovecot with mariadb

vim /etc/dovecot/conf.d/auth-sql.conf.ext

# Authentication for SQL users. Included from 10-auth.conf.
#
# &amp;lt;doc/wiki/AuthDatabase.SQL.txt&amp;gt;

passdb {
  driver = sql

  # Path for SQL configuration file, see example-config/dovecot-sql.conf.ext
  args = /etc/dovecot/conf.d/dovecot-mysql.conf.ext
}

# &amp;quot;prefetch&amp;quot; user database means that the passdb already provided the
# needed information and there's no need to do a separate userdb lookup.
# &amp;lt;doc/wiki/UserDatabase.Prefetch.txt&amp;gt;
#userdb {
#  driver = prefetch
#}

userdb {
  driver = sql
  args = /etc/dovecot/conf.d/dovecot-mysql.conf.ext
}

# If you don't have any user-specific settings, you can avoid the user_query
# by using userdb static instead of userdb sql, for example:
# &amp;lt;doc/wiki/UserDatabase.Static.txt&amp;gt;
#userdb {
  #driver = static
  #args = uid=vmail gid=vmail home=/var/vmail/%u
#}

vim /etc/dovecot/conf.d/10-auth.conf

#!include auth-system.conf.ext
!include auth-sql.conf.ext

at this point we have a basical dovecot service running

systemctl restart dovecot.service

but we dont enable service if we restart our computer we don’t have dovecot enabled

systemctl enable dovecot.service

step 28: open Dovecot ports in firewall

dovecot is ready but we need to open ports in firewall

firewall-cmd –permanent –zone=public –add-service=pop3s

firewall-cmd –permanent –zone=public –add-port=110/tcp

firewall-cmd –permanent –zone=public –add-service=imaps

firewall-cmd –permanent –zone=public –add-port=143/tcp
firewall-cmd –reload

step 29: configure postfix smtp authentcation

when we configured dovecot we configure embebed sasl authentication, now we will integrate dovecot auth in postfix

we need to append these lines to our main.cf file

vim  /etc/postfix/main.cf

smtpd_sasl_type = dovecot
smtpd_sasl_path = private/auth
smtpd_sasl_auth_enable = yes
smtpd_relay_restrictions = permit_mynetworks, permit_sasl_authenticated, reject_unauth_destination

step 30: avoid plain logins in smtp connections

Security parameters appear to be well configured, but if we leave configuration in hand of users, they will try to use plain login, this is easily to sniff we avoid these options

vim /etc/postfix/main.cf

and append these lines

#disallow plain login
smtpd_tls_auth_only = yes

 

step 31: enable smtp submission port

At this moment we know that our smtp service should run only in port 25, unfortunately some Internet Service Providers decided that the best way to  stop spam is deny conections to smtp port. If you leave only port 25 opened some users can’t send email because their ISP deny connections. To solve this problem we can enable  SMTP submission port, 587. and forget phone calls from users complaining about that they can’t send email from cafes, or their homes.

we need to modify file  /etc/postfix/master.cf

vim /etc/postfix/master.cf

and remove comments in these lines

submission inet n       -       n       -       -       smtpd
  -o syslog_name=postfix/submission
  -o smtpd_tls_security_level=encrypt
  -o smtpd_sasl_auth_enable=yes
  -o smtpd_reject_unlisted_recipient=no
  -o smtpd_client_restrictions=$mua_client_restrictions
  -o smtpd_helo_restrictions=$mua_helo_restrictions
  -o smtpd_sender_restrictions=$mua_sender_restrictions
  -o smtpd_recipient_restrictions=permit_sasl_authenticated,reject
  -o milter_macro_daemon_name=ORIGINATING

we also need to define mua variables

vim /etc/postfix/main.cf


mua_client_restrictions = permit_sasl_authenticated,reject
mua_helo_restrictions = permit_sasl_authenticated,reject
mua_sender_restrictions = permit_sasl_authenticated,reject

now we will restart postifx service

 systemctl restart postfix.service

and we should get one output like these is /var/log/maillog

tail -f /var/log/maillog

May 10 16:08:51 mail postfix/postfix-script[13671]: stopping the Postfix mail system
May 10 16:08:51 mail postfix/master[3802]: terminating on signal 15
May 10 16:08:52 mail postfix/postfix-script[13752]: starting the Postfix mail system
May 10 16:08:52 mail postfix/master[13754]: daemon started -- version 2.10.1, configuration /etc/postfix

we need to open smtp submission port too

firewall-cmd –permanent –zone=public –add-port=587/tcp

firewall-cmd –reload

 Step 32: Test configuration against a mail client

In my case I will use Thunderbird, there are a lot of mail clients. If you use self signed certificates you should search in google, duck duck go or another web searcher how it’s the client behavior.

Configuring this data is easy.

  • all your servers will be your fully qualified domain name,
  • your username will be the email account
  • and the password will be that you put in postfixadmin

check and send some emails between accounts, don’t send to hotmail or gmail probabily it will appear in spam.

it should work.

if doesn’t work check /var/log/maillog output or leave a comment

step 33:link amavisd-new with mariadb database

amavisd neew to know what domains is serving we need to link Mariadb database with amavisd-new.

In order to maintain a security lever we will create a view and a new user that only can access to this view.

we will append these lines in amavisd config file

vim /etc/amavisd/amavisd.conf

@lookup_sql_dsn = (
    ['DBI:mysql:database=postfix;host=127.0.0.1;port=3306', 'postfixread', 'yourpassword']
);

$sql_select_policy =  'SELECT &amp;quot;Y&amp;quot; AS local FROM domain WHERE CONCAT(&amp;quot;@&amp;quot;, domain) IN (%k)';

amavisd default config will discard spam mail we want to changue this behavior

vim /etc/amavisd/amavisd.conf

and edit final_spam_destiny variable

$final_spam_destiny       = D_PASS;  #!!!  D_DISCARD / D_REJECT

now amavisd will deliver spam messages to inbox and knows what domains are hosted with postfixadmin.

Amavisd sends a copy of email to spamassasin and reads spamassasin results, this means that  if spamassasin modifies email subject these will not appears in our detected email. we should modify several parameters in amavis config file to always write spamassasing analysis results.

vim /etc/amavisd/amavisd.conf

just modify these values

$sa_tag_level_deflt  = -9999;  # add spam info headers if at, or above that level
$sa_tag2_level_deflt = 6.2;  # add 'spam detected' headers at that level
$sa_kill_level_deflt = 6.9;  # triggers spam evasive actions (e.g. blocks mail)
$sa_dsn_cutoff_level = 10;   # spam level beyond which a DSN is not sent
$sa_crediblefrom_dsn_cutoff_level = 18; # likewise, but for a likely valid From

We will  reconfigure some parameters in spamassasin

 

step 34: Configure spamassasin

vim /etc/mail/spamassassin/local.cf

should be like these

required_hits 5.0
report_safe 0
required_score 5
remove_header ham Status
remove_header ham Level

We can link spamassasing with a mariadb database that force us to create a database

mysql -u root -p

CREATE DATABASE mail_spamassassin;
CREATE USER 'spamassassin'@'localhost' IDENTIFIED BY 'new_password';
GRANT ALL PRIVILEGES ON `mail_spamassassin` . * TO 'spamassassin'@'localhost';
FLUSH PRIVILEGES;
quit;

we need to create several tables

mysql -u root -p mail_spamassassin


CREATE TABLE bayes_expire (
  id int(11) NOT NULL default '0',
  runtime int(11) NOT NULL default '0',
  KEY bayes_expire_idx1 (id)
) ENGINE=InnoDB;

CREATE TABLE bayes_global_vars (
  variable varchar(30) NOT NULL default '',
  value varchar(200) NOT NULL default '',
  PRIMARY KEY  (variable)
) ENGINE=InnoDB;

INSERT INTO bayes_global_vars VALUES ('VERSION','3');

CREATE TABLE bayes_seen (
  id int(11) NOT NULL default '0',
  msgid varchar(200) binary NOT NULL default '',
  flag char(1) NOT NULL default '',
  PRIMARY KEY  (id,msgid)
) ENGINE=InnoDB;

CREATE TABLE bayes_token (
  id int(11) NOT NULL default '0',
  token binary(5) NOT NULL default '',
  spam_count int(11) NOT NULL default '0',
  ham_count int(11) NOT NULL default '0',
  atime int(11) NOT NULL default '0',
  PRIMARY KEY  (id, token),
  INDEX bayes_token_idx1 (id, atime)
) ENGINE=InnoDB;

CREATE TABLE bayes_vars (
  id int(11) NOT NULL AUTO_INCREMENT,
  username varchar(200) NOT NULL default '',
  spam_count int(11) NOT NULL default '0',
  ham_count int(11) NOT NULL default '0',
  token_count int(11) NOT NULL default '0',
  last_expire int(11) NOT NULL default '0',
  last_atime_delta int(11) NOT NULL default '0',
  last_expire_reduce int(11) NOT NULL default '0',
  oldest_token_age int(11) NOT NULL default '2147483647',
  newest_token_age int(11) NOT NULL default '0',
  PRIMARY KEY  (id),
  UNIQUE bayes_vars_idx1 (username)
) ENGINE=InnoDB;

CREATE TABLE awl (
  username varchar(100) NOT NULL default '',
  email varbinary(255) NOT NULL default '',
  ip varchar(40) NOT NULL default '',
  count int(11) NOT NULL default '0',
  totscore float NOT NULL default '0',
  signedby varchar(255) NOT NULL default '',
  PRIMARY KEY (username,email,signedby,ip)
) ENGINE=InnoDB;

quit;

once when we have created these tables we need to link spamassasin with these tables

Auto-Whitelists linking

we need to enable autowhitelist check in spamassasin config

 vim /etc/mail/spamassassin/v310.pre

edit line 45 removing comment

# AWL - do auto-whitelist checks
#
loadplugin Mail::SpamAssassin::Plugin::AWL

and configure database access

vim /etc/mail/spamassassin/auto-whitelist.cf

auto_whitelist_factory Mail::SpamAssassin::SQLBasedAddrList

user_awl_dsn                 DBI:mysql:mail_spamassassin:localhost
user_awl_sql_username        spamassassin
user_awl_sql_password        yourpassword

use_auto_whitelist 1

Bayesian Storage Module

vim /etc/mail/spamassassin/bayesian-storage.cf

bayes_store_module                 Mail::SpamAssassin::BayesStore::SQL

bayes_sql_dsn                 DBI:mysql:mail_spamassassin:localhost
bayes_sql_username            spamassassin
bayes_sql_password            yourpassword

use_bayes          1
bayes_auto_learn   1
bayes_auto_expire  1

at this point we will restart spamassassin service

systemctl restart spamassassin.service

we should see a log like this in /var/log/maillog

May 11 12:18:30 mail spamd[23346]: logger: removing stderr method
May 11 12:18:36 mail spamd[23348]: spamd: server started on IO::Socket::IP [127.0.0.1]:783, IO::Socket::IP [::1]:783 (running version 3.4.0)
May 11 12:18:36 mail spamd[23348]: spamd: server pid: 23348
May 11 12:18:36 mail spamd[23348]: spamd: server successfully spawned child process, pid 23354
May 11 12:18:36 mail spamd[23348]: spamd: server successfully spawned child process, pid 23355
May 11 12:18:36 mail spamd[23348]: prefork: child states: IS
May 11 12:18:36 mail spamd[23348]: prefork: child states: II

Now we will give a spam message to spamassassin to start learning

 sa-learn –spam /usr/share/doc/spamassassin-3.4.0/sample-spam.txt

output should be like these

Learned tokens from 1 message(s) (1 message(s) examined)

this acction will add some data into database if we loged on mariadb

and execute a

select * from mail_spamassassin.bayes_vars;

we should see one record.

step 35: Sending spam to a dedicated folder in each account

When someone send spam it will appear in our inbox with text ***SPAM***  in the subject.

inbox spam detectedif stats about 99% of received email are right, that means that we will get 99 ***Spam*** mails before get a valid one. Leaving a lot of Spam a inbox could be irritating for our users.

One smart solution used by providers like Google, Microsoft or Yahoo is deliver spam in a dedicated folder.  Probably your users are familiarized with these behavior.

In our actual configuration deliver email to mailbox is managed by postfix virtual domain agent delivery VIRTUAL(8) , virtual agent can’t deliver email to spam folder. We need to use one alternative like Dovecot.

With dovecot acting as local delivery we can use sieve  filters to manage messages location.

For local delivery Dovecot offers two options, LDA and LMTP

  • LDA works like a binary command, each time that postfix sends a email lda deliver is called.
  • LMTP is like a long-running process started by Dovecot.

obviously  LMTP is better and is the option that we should use.

We need to change delivery agent in postfix. This requires some configuration changes in dovecot.

 

vim /etc/dovecot/conf.d/10-master.conf

and fill lmtp settings like these

service lmtp {
   unix_listener /var/spool/postfix/private/dovecot-lmtp {
     group = postfix
     mode = 0600
     user = postfix
    }
  unix_listener lmtp {
    #mode = 0666
   }

  # Create inet listener only if you can't use the above UNIX socket
  #inet_listener lmtp {
    # Avoid making LMTP visible for the entire internet
    #address =
    #port =
  #}
}

next step is configure lmtp protocol itself

 vim /etc/dovecot/conf.d/20-lmtp.conf

here you will need to give a postmaster address

protocol lmtp {
postmaster_address = postmaster@yourdomain.com
}

finally we will enable lmtp protocol

vim /etc/dovecot/dovecot.conf


protocols = imap pop3  lmtp

our configuration is ready we should restart dovecot service to apply configuration modifications

systemctl restart dovecot.service

now postfix need to use dovecot lda instead virtual

append these lines in postfix main config file

vim /etc/postfix/main.cf

#use dovecot lmtp as virtual transport
virtual_transport = lmtp:unix:private/dovecot-lmtp

after these modifications we can restart postfix service

systemctl restart postfix.service

at this point send and email from one virtual account to another, it should work without problems.

in /var/log/maillog it should appear a line like this

dovecot: lmtp(3762, youruser@yourdomain.com): XXXXXXXXXXXXXXXXXX: msgid=XXXXXXXXXXX.YYYYYYYYY@yourdomain.com: saved mail to INBOX

Now dovecot is our local delivery agent

before enable sieve plugin we need to ensure that every IMAP user will have a spam folder

vim /etc/dovecot/conf.d/15-mailboxes.conf

We will create an typical set of folders

##
## Mailbox definitions
##

# NOTE: Assumes &amp;amp;quot;namespace inbox&amp;amp;quot; has been defined in 10-mail.conf.
namespace inbox {

  #mailbox name {
    # auto=create will automatically create this mailbox.
    # auto=subscribe will both create and subscribe to the mailbox.
    #auto = no

    # Space separated list of IMAP SPECIAL-USE attributes as specified by
    # RFC 6154: \All \Archive \Drafts \Flagged \Junk \Sent \Trash
    #special_use =
  #}

  # These mailboxes are widely used and could perhaps be created automatically:
  mailbox Drafts {
    auto = subscribe
    special_use = \Drafts
  }
  mailbox Junk {
    auto = subscribe
    special_use = \Junk
  }
  mailbox Trash {
    auto = subscribe
    special_use = \Trash
  }

  # For \Sent mailboxes there are two widely used names. We'll mark both of
  # them as \Sent. User typically deletes one of them if duplicates are created.
  mailbox Sent {
    auto=subscribe
    special_use = \Sent
  }

  # If you have a virtual &amp;amp;quot;All messages&amp;amp;quot; mailbox:
  #mailbox virtual/All {
  #  special_use = \All
  #}

  # If you have a virtual &amp;amp;quot;Flagged&amp;amp;quot; mailbox:
  #mailbox virtual/Flagged {
  #  special_use = \Flagged
  #}
}

restart dovecot service

systemctl restart dovecot.service

now if we open our mail client, thunderbird, we should see a list of new folders

imap folders

after all those steps we can enable sieve plugin

we will create a sieve global filter

cd /var/vmail

mkdir sieve

cd sieve

vim globalfilter.sieve

fill file with this content

require &amp;quot;fileinto&amp;quot;;
  if exists &amp;quot;X-Spam-Flag&amp;quot; {
          if header :contains &amp;quot;X-Spam-Flag&amp;quot; &amp;quot;NO&amp;quot; {
          } else {
          fileinto &amp;quot;Junk&amp;quot;;      
          stop;
      }
  }
  if header :contains &amp;quot;subject&amp;quot; [&amp;quot;***SPAM***&amp;quot;] {
    fileinto &amp;quot;Junk&amp;quot;;
    stop;
  }

chown -R vmail:mail /var/vmail/sieve

we need to enable sieve services in dovecot

vim /etc/dovecot/conf.d/20-managesieve.conf


##
## ManageSieve specific settings
##

# Uncomment to enable managesieve protocol:
#protocols = $protocols sieve

# Service definitions

service managesieve-login {
  inet_listener sieve {
    port = 4190
  }

  #inet_listener sieve_deprecated {
  #  port = 2000
  #}

  # Number of connections to handle before starting a new process. Typically
  # the only useful values are 0 (unlimited) or 1. 1 is more secure, but 0
  # is faster. &amp;lt;doc/wiki/LoginProcess.txt&amp;gt;

  service_count = 1

  # Number of processes to always keep waiting for more connections.
  process_min_avail = 0

  # If you set service_count=0, you probably need to grow this.
  vsz_limit = 64M
}

service managesieve {
  # Max. number of ManageSieve processes (connections)
  #process_limit = 1024
}

# Service configuration

protocol sieve {
  # Maximum ManageSieve command line length in bytes. ManageSieve usually does
  # not involve overly long command lines, so this setting will not normally
  # need adjustment
  managesieve_max_line_length = 65536

  # Maximum number of ManageSieve connections allowed for a user from each IP
  # address.
  # NOTE: The username is compared case-sensitively.
  #mail_max_userip_connections = 10

  # Space separated list of plugins to load (none known to be useful so far).
  # Do NOT try to load IMAP plugins here.
  #mail_plugins =

  # MANAGESIEVE logout format string:
  #  %i - total number of bytes read from client
  #  %o - total number of bytes sent to client
  #managesieve_logout_format = bytes=%i/%o

  # To fool ManageSieve clients that are focused on CMU's timesieved you can
  # specify the IMPLEMENTATION capability that Dovecot reports to clients.
  # For example: 'Cyrus timsieved v2.2.13'
  managesieve_implementation_string = Dovecot Pigeonhole

  # Explicitly specify the SIEVE and NOTIFY capability reported by the server
  # before login. If left unassigned these will be reported dynamically
  # according to what the Sieve interpreter supports by default (after login
  # this may differ depending on the user).
  #managesieve_sieve_capability =
  #managesieve_notify_capability =

  # The maximum number of compile errors that are returned to the client upon
  # script upload or script verification.
  managesieve_max_compile_errors = 5

  # Refer to 90-sieve.conf for script quota configuration and configuration of
  # Sieve execution limits.
}

vim /etc/dovecot/conf.d/90-plugin.conf


plugin {
  #setting_name = value
  sieve_global_path = /var/vmail/sieve/globalfilter.sieve
  sieve_max_script_size = 1M
}

local delivery should use sieve

vim /etc/dovecot/conf.d/20-lmtp.conf


protocol lmtp {
  # Space separated list of plugins to load (default is global mail_plugins).
  postmaster_address =  postmaster@yourdomain.com
  mail_plugins = $mail_plugins sieve
}

at this point we should restart dovecot to apply these changues

systemctl restart dovecot.service

check service status if something is bad

systemctl status dovecot.service

now you can send a gtube string to yourself, it should appear in junk folder


May 12 12:01:07 mail postfix/submission/smtpd[12410]: connect from yourcomputer[yourip]
May 12 12:01:07 mail postfix/submission/smtpd[12410]: 8D6F310095C: client=XEON.cafeingles[yourip], sasl_method=PLAIN, sasl_username=destination@domain.com
May 12 12:01:07 mail postfix/cleanup[12420]: 8D6F310095C: message-id=&amp;lt;5551CF63.8070307@yourfqdnsourcedomain.com&amp;gt;
May 12 12:01:07 mail postfix/qmgr[1977]: 8D6F310095C: from=&amp;lt;destination@domain.com&amp;gt;, size=629, nrcpt=1 (queue active)
May 12 12:01:07 mail postfix/submission/smtpd[12410]: disconnect from XEON.cafeingles[yourip]
May 12 12:01:07 mail amavis[2763]: (02763-08) NOTICE: reconnecting in response to: err=2006, HY000, DBD::mysql::st execute failed: MySQL server has gone away at (eval 129) line 172.
May 12 12:01:08 mail postfix/smtpd[12427]: connect from localhost[127.0.0.1]
May 12 12:01:08 mail postfix/smtpd[12427]: 74AB1100960: client=localhost[127.0.0.1]
May 12 12:01:08 mail postfix/cleanup[12420]: 74AB1100960: message-id=&amp;lt;5551CF63.8070307@yourfqdnsourcedomain.com&amp;gt;
May 12 12:01:08 mail postfix/qmgr[1977]: 74AB1100960: from=&amp;lt;destination@domain.com&amp;gt;, size=1395, nrcpt=1 (queue active)
May 12 12:01:08 mail postfix/smtpd[12427]: disconnect from localhost[127.0.0.1]
May 12 12:01:08 mail amavis[2763]: (02763-08) Passed SPAM {RelayedTaggedInternal,Quarantined}, MYNETS LOCAL [yourip]:36125 &amp;lt;destination@domain.com&amp;gt; -&amp;gt; &amp;lt;destination@domain.com&amp;gt;, Queue-ID: 8D6F310095C, Message-ID: &amp;lt;5551CF63.8070307@yourfqdnsourcedomain.com&amp;gt;, mail_id: PxRBJkjebYHJ, Hits: 999.001, size: 629, queued_as: 74AB1100960, 885 ms
May 12 12:01:08 mail postfix/smtp[12421]: 8D6F310095C: to=&amp;lt;destination@domain.com&amp;gt;, relay=127.0.0.1[127.0.0.1]:10024, delay=0.99, delays=0.04/0.03/0.02/0.89, dsn=2.0.0, status=sent (250 2.0.0 from MTA(smtp:[127.0.0.1]:10025): 250 2.0.0 Ok: queued as 74AB1100960)
May 12 12:01:08 mail postfix/qmgr[1977]: 8D6F310095C: removed
May 12 12:01:08 mail dovecot: lmtp(12429): Connect from local
May 12 12:01:08 mail dovecot: lmtp(12429, destination@domain.com): UOvhIGTPUVWNMAAAiaI7Ow: sieve: msgid=&amp;lt;5551CF63.8070307@yourfqdnsourcedomain.com&amp;gt;: stored mail into mailbox 'Junk'
May 12 12:01:08 mail postfix/lmtp[12428]: 74AB1100960: to=&amp;lt;destination@domain.com&amp;gt;, relay=mail.yourfqdn.com[private/dovecot-lmtp], delay=0.1, delays=0.01/0.03/0.03/0.02, dsn=2.0.0, status=sent (250 2.0.0 &amp;lt;destination@domain.com&amp;gt; UOvhIGTPUVWNMAAAiaI7Ow Saved)
May 12 12:01:08 mail postfix/qmgr[1977]: 74AB1100960: removed
May 12 12:01:08 mail dovecot: lmtp(12429): Disconnect from local: Successful quit

and your mail client should show a new mail in junk folder

spam in junk folder

Step 36: Training spamassasin

Our detected spam travels to junk folder, but if forget to train spamassasin we will get a lot of spam in our Inbox folder or false positives in our Junk folder.

To avoid this we can launch sa-learn command but do this by hand is a tedious work. Good sysadmins automatize these task, I make a small script in python, my first python script, to launch sa-learn automatically.

This script will scan each mailbox and check if user has created an extra imap folder, to ignore pop3 only users, adding all mail in junk folder as spam and rest of folders as ham mail.

I created a project in github if you want to colaborate I will accept all help

https://github.com/luzemail/spamAssasingTraining

to add this script to our server we will execute a set of commands

 

cd /var/vmail

wget https://raw.githubusercontent.com/luzemail/spamAssasingTraining/master/trainspamassasin.py

chmod +x trainspamassasin.py

now we will add a crontab line to execute this script all days

vim /etc/crontab

add this line


0  0  *  *  * vmail       /var/vmail/trainspamassasin.py

adding a little of python magic helps to forget about what is learning spamassasin.

A small trick to start this is tell to our user that they should create a folder called nospam or whatever you prefer and recommend  users to move false posivitives to this folder and undetected spam to junk folder.

This script scans all mail folders in account and ads all mail as ham except junk folder which is added as spam. You should need to change this behavior feel free to modify this script and colaborate.

Step 37:Enabling greylisting (optional)

One technique to avoid spam is use a greylisting, basically when we receive a mail for first time our mailserver will answer with an internal error forcing to sender, if senders server is well configures it will try in a few minutes to deliver message again and it will be acepted. Spamers doesn’t resend emails they simply forget our mailserver.

So using a greylisting adds delay receiving mails or they can be undelivered because another server isn’t well configured.

first we need to install postgrey

yum install postgrey

postgrey works as a daemon we need to activate it

systemctl start postgrey.service

and enable service

systemctl enable postgrey.service

we can check postgrey service status running

systemctl status postgrey.service

we should get an output like this


postgrey.service - Postfix Greylisting Service
Loaded: loaded (/usr/lib/systemd/system/postgrey.service; disabled)
Active: active (running) since Sat 2015-05-23 15:35:24 CEST; 13s ago
Docs: man:postgrey(8)
Process: 8176 ExecStart=/usr/sbin/postgrey --unix=/var/spool/postfix/postgrey/socket --pidfile=/var/run/postgrey.pid --group=postgrey --user=postgrey --greylist-text=Greylisted for %%s seconds --daemonize $POSTGREY_OPTS (code=exited, status=0/SUCCESS)
Process: 8173 ExecStartPre=/bin/rm -f /var/run/postgrey.pid (code=exited, status=0/SUCCESS)
Main PID: 8177 (/usr/sbin/postg)
CGroup: /system.slice/postgrey.service
└─8177 /usr/sbin/postgrey --unix=/var/spool/postfix/postgrey/socket --pidfile=/var/run/postgrey.pid --group=postgrey --user=postgrey --greylist-text=Greylisted for %s seconds --daemonize --delay=60...

now we need to link postgrey with postfix

just edit postfix config file

vim /etc/postfix/main.cf

and append these lines


smtpd_recipient_restrictions =
permit_mynetworks,
permit_sasl_authenticated,
check_policy_service unix:/var/spool/postfix/postgrey/socket

and restart  postfix service

 systemctl restart postfix.service

if you need to modify postgrey behavior you can use these files:

  • /etc/postfix/postgrey_whitelist_clients
  • /etc/postfix/postgrey_whitelist_recipients
  • /etc/postfix/postgrey_whitelist_clients.local

 

Step 38: Configuring smtpd_recipient_restrictions

Like greylisting we can add some restrictions when receiving email to avoid receive unsolicited mail.

we can add these restriction to our smtpd_sender_restrictions line

  • reject_unknown_address –> mails without from
  • reject_unknown_sender_domain –> no know sender
  • reject_invalid_hostname –> when a server make helo with a malformed hostname
  • reject_unknown_recipient_domain –> If postfix is not final distination for recipient domain
  • reject_unauth_pipelining –> stops mail from bulk mail software that doesn’t comply ESMTP command pipelining

finally we can use RBL’s, Real time Blackhole lists is a list of know spammers ips that will be rejected by postfix

command is basically


reject_rbl_client server1,

reject_rbl_client server2,

....,

reject_rbl_client servern,

permit

this will help you to stop a lot of spam and free system resources.

here you get a list of rbls http://en.wikipedia.org/wiki/Comparison_of_DNS_blacklists

here you have an example configuration

vim /etc/postfix/main.cf


smtpd_recipient_restrictions =
reject_invalid_hostname,
reject_unknown_recipient_domain,
reject_unauth_pipelining,
permit_mynetworks,
reject_unauth_destination,
permit_sasl_authenticated,
check_policy_service unix:/var/spool/postfix/postgrey/socket,
reject_rbl_client zen.spamhaus.org,
reject_rbl_client bl.spamcop.net,
reject_rbl_client dnsbl.sorbs.net,
reject_rbl_client cbl.abuseat.org,
reject_rbl_client b.barracudacentral.org,
reject_rbl_client dnsbl-1.uceprotect.net,
permit

and finally restart postfix

 systemctl restart postfix.service

Step 39: Enabling spf

SPF (Sender Policy Framework) is basically a TXT dns record that indicates what ips and/or domains are authorized to send email.

A minimal spf record should be like this


spf=v1 mx -all

if your webserver send emails too


spf=v1 a mx -all

if you want to add an ip


spf=v1 a mx ip4:255.255.255.255 -all

 

in function of what servers will be generating email you should configure your spf records.

Maintain a spf record in your domains is important to avoid that spammers use your domain as as sender part of domain

Step 40: Enabling DKIM

DKIM, DomainKeys Identified Mail, helps to verify that a received email is sender from a valid mail server. Basically is a digital signature that can be verified using a public key published into senders domain dns record.

We need to generate a public and private key for each domain.

basically for each domain we need to execute a command like this

opendkim-genkey –bits=4096 –domain=example.com –selector=example.com –restrict

be careful with your dns provider 4096 bits could generate a long public key that couldn’t fit in your txt box, in this case you can reduce your key length to 2048 or changue dns provider

for generate these keys we should follow these steps:

go to opendkim keys dir

cd /etc/opendkim/keys

generate one keypair for each domain

opendkim-genkey –bits=4096 –domain=example.com –selector=example.com –restrict

this command will create two files

  • domain.com.private your private key
  • domain.com.txt contains public part in dns record format

Note: your should renew your keys every year.

we will to changue private keys ownership

 chown opendkim:opendkim /etc/opendkim/keys/*.private

we need to configure opendkim edit config file

vim /etc/opendkim.conf

options should be like these

PidFile    /var/run/opendkim/opendkim.pid
Mode    sv
Syslog    yes
SyslogSuccess    yes
LogWhy    yes
UserID    opendkim:opendkim
Socket    inet:8891@localhost
Umask    002
Canonicalization    relaxed/relaxed
Selector    default
MinimumKeyBits 1024
KeyTable    refile:/etc/opendkim/KeyTable
SigningTable    refile:/etc/opendkim/SigningTable
ExternalIgnoreList    refile:/etc/opendkim/TrustedHosts
InternalHosts    refile:/etc/opendkim/TrustedHosts

OpenDKim needs a list of host whose mail should signed by Opendkim-

vim /etc/opendkim/TrustedHosts

content should be like these


127.0.0.1
::1
mail.yourdomain.com

mail.yourdomain.com should be your hostname

all mail originated from these host will be signed otherwise is ignored, if you have a relay server you should add these ip.

We will host several domains in our configuration, OpenDkim needs to know a list of domains and keys to sign.

vim /etc/opendkim/KeyTable

we need to link each domain with one key adding lines like this


default._domainkey.domain1.com domain1.com:default:/etc/opendkim/keys/domain1.com.private
default._domainkey.domain2.com domain2.com:default:/etc/opendkim/keys/domain2.com.private

using these list opendkim knows relations between domains and private keys.

now opendkim needs to know relation between mail adress and domains whe should configure SigningTable file

vim /etc/opendkim/SigningTable

if we want to sign all email address we shoild add a line like these


*@domain1.com default._domainkey.domain1.com

*@domain2.com default._domainkey.domain2.com

complete with all your domains and restart opendkim

systemctl restart opendkim.service

now you can check opendkim service status

 systemctl status opendkim.service

output should be like these


opendkim.service - DomainKeys Identified Mail (DKIM) Milter
Loaded: loaded (/usr/lib/systemd/system/opendkim.service; enabled)
Active: active (running) since Wed 2015-05-27 16:26:39 CEST; 1s ago
Docs: man:opendkim(8)
man:opendkim.conf(5)
man:opendkim-genkey(8)
man:opendkim-genzone(8)
man:opendkim-testadsp(8)
man:opendkim-testkey
http://www.opendkim.org/docs.html
Process: 23824 ExecStart=/usr/sbin/opendkim $OPTIONS (code=exited, status=0/SUCCESS)
Main PID: 23825 (opendkim)
CGroup: /system.slice/opendkim.service
└─23825 /usr/sbin/opendkim -x /etc/opendkim.conf -P /var/run/opendkim/opendkim.pid

May 27 16:26:39 mail.yourdomain.com systemd[1]: Started DomainKeys Identified Mail (DKIM) Milter.
May 27 16:26:39 mail.yourdomain.com opendkim[23825]: OpenDKIM Filter v2.10.1 starting (args: -x /etc/opendkim.conf -P /var/run/opendkim/opendkim.pid)

At this point we have a opendkim daemon working we need to integrate opendkim with postfix

we need to open postfix main config file

vim /etc/postfix/main.cf

and finally append these lines


milter_default_action = accept
smtpd_milters = inet:127.0.0.1:8891

and restart postfix to see these changues applied

systemctl restart postfix.service

Finally the most important step is publish your public keys in each domain

for example for a file example.com.txt

whit these content


example.com._domainkey    IN    TXT    ( &amp;quot;v=DKIM1; k=rsa; s=email; &amp;quot;
&amp;quot;p=MIICIjANBgkqhkiG9w0BAQEFAAOCAg8AMIICCgKCAgEAzGqMtyZwjFzNsvFVSnsPvyHcsAGpqglHcKtRSIGKyzbAigp18LARojk5UlTAHmED46THNbo8q6IS/fysaGBBR97oZLK2/8Vl6FOc0hdU1alEaAr/MpI+0MquCqjaAFdVWKHtGRthCzJ0HuDqbQFBnc3eUOe8RxkYWwxyKs+Tze/FCQ/mso/Gm/Zp/z7v8jTRbaIZtKRB+1oBrc&amp;quot;
&amp;quot;L2WuFEOkZyxCEq0gYLNV2AYcfIdvBXqHsHLeeEZMEbxIHOQGg3fINd3bbP2hxWtlnCrIGFQxdkOH4hx75wfZ+QRWh0d7jmW4c0Jnwvw0HLIJSzfS1kOUCPSq+MR7h4bT17sfWMXSvwqWca1R0eVRZdkuuBBeK5897vvRCA/44WMhv2GeWM6uHrRLy8Z8CAoCVd4FrZ6UQ+eQ2SjJaObInWbXC0/VRNHLRHVqW3pZROH3tYWAD39EUKpAWO&amp;quot;
&amp;quot;vr6YfwD/7PeM/283LLDuQceqIVg4kYcNeZR9iL65sLXWkHPb8rJeGFqQhUC+Cvm1HkhLbm5m/OHl41EF+dfLDT+c8EpCT3khSebKvwHFbd2l6XQhy+zQSvQtPSgtWJ2mgq4FIHFBJFdQRUv/KQdhDDapYkcM80DyyRnzXv4erfNcC//LaI4LmJBd36KTNKWUbrBibxOWye3ZheiaPGYrdwrp/X0rPXnnju8CAwEAAQ==&amp;quot; )  ; ----- DKIM key example.com for example.com

you need to add a _domainkey txt record whit this text


v=DKIM1; k=rsa; s=email; p=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

Step 41: Configuring DMARC records for each domain

Basically Dmarc is a method that joins SPF and DKIM defining what to do when a received mail doesn’t pass DKIM or SPF

we need to create a DNS TXT record like _dmarc.your_domain.com

with this text

v=DMARC1 p=pvalue optinalparameter1=value optinalparameter2=value …

Mandatory Parameters

Policy of domain
p= pvalues
Defines policy of domain.

pvalues are:

  • none ->The Domain Owner requests no specific action be taken regarding delivery of messages
  • quarantine -> if check dkim and/or spf fails, Depending on the capabilities of the Mail Receiver, this can mean “place into spam folder”, “scrutinize with additional intensity”, and/or “flag as suspicious”.
  • reject  -> check dkim and/or spf fails Mail Receiver will reject mail during smtp transaction

Optional Parameters

Policy of Subdomains

sp= spvalues

spvalues are:

  • none ->The Domain Owner requests no specific action be taken regarding delivery of messages
  • quarantine -> if check dkim and/or spf fails, Depending on the capabilities of the Mail Receiver, this can mean “place into spam folder”, “scrutinize with additional intensity”, and/or “flag as suspicious”.
  • reject  -> check dkim and/or spf fails Mail Receiver will reject mail during smtp transaction

DKIM identifier alignment

adkim= dkimvalues

dkimvalues are:

s -> strict, sender domain name must be the same that d=name in DKIM mail headers else fails
r-> relaxed, if sender domain is a subdomain will pass

if omitted adkim tag in domain txt default value is r (relaxed)

SPF-authenticated Identifiers

aspf= aspfvalues

aspfvalues are:

s -> strict, MAIL FROM command in SMTP and from:header in email must mach
r-> relaxed, if sender domain is a subdomain will pass

if omitted aspf tag in domain txt default value is r (relaxed)

Percentage of messages where DMARC mechanism is to be applied

pct=pctvalue

pctvalue= a number between 0 and 100

Interval between Aggregaye reports
ri = rivalue

rivalue= number of seconds between reports default 86400 (24 hours)

Reporting URL of aggregate reports

rua:mailto:address@domain.com

Each server will send an aggregate feedback to this adress

if domain lies outside sending zone you must validate, use a web searcher

Reporting URL of aggregate feedback

ruf:mailto:address@domain.com

Used for forensic reports about messagest that fail spf and/or dkim evaluation

if domain lies outside sending zone you must validate, use a web searcher

here you have a example of dmarc record


v=DMARC1; p=quarantine; pct=100 rua=mailto:dmarc.rua@customddomain.com

DMARC is not a point and shoot implementation you should follow a sequence like this in function of your mail volume:

  1.     Monitor all.
  2.    Quarantine 1%.
  3.     Quarantine 5%.
  4.     Quarantine 10%.
  5.   Quarantine 25%.
  6.   Quarantine 50%.
  7.   Quarantine all.
  8.   Reject 1%.
  9.   Reject 5%.
  10.   Reject 10%.
  11.   Reject 25%.
  12.   Reject 50%.
  13.   Reject all.

 

Step 42: Enabling DMARC suport in postfix

Receive reports from DMARC helps to verify our configurations, configure a mailserver that send DMARC reports to another sysadmins makes you a good sysadmin and a better person :-P.

To make this we need to configure openDMARC

Actually in centos 7 openDMARC needs libspf2 that is avaliable in testing repo.

if you try to install opendmarc running this command

yum install opendmarc

and you get an output like these


---&amp;gt; Package libopendmarc.x86_64 0:1.3.1-4.el7 will be updated
---&amp;gt; Package libopendmarc.x86_64 0:1.3.1-13.el7 will be an update
--&amp;gt; Processing Dependency: libspf2.so.2()(64bit) for package: libopendmarc-1.3.1-13.el7.x86_64
---&amp;gt; Package opendmarc.x86_64 0:1.3.1-4.el7 will be updated
---&amp;gt; Package opendmarc.x86_64 0:1.3.1-13.el7 will be an update
--&amp;gt; Processing Dependency: libspf2.so.2()(64bit) for package: opendmarc-1.3.1-13.el7.x86_64
--&amp;gt; Finished Dependency Resolution
Error: Package: libopendmarc-1.3.1-13.el7.x86_64 (epel)
Requires: libspf2.so.2()(64bit)
Error: Package: opendmarc-1.3.1-13.el7.x86_64 (epel)
Requires: libspf2.so.2()(64bit)
You could try using --skip-broken to work around the problem
You could try running: rpm -Va --nofiles --nodigest

run this command

yum install opendmarc –enablerepo=epel-testing

OpenDMARK will need a MariaDB database  in order to store all info needed to generate reports.

We will create this database as first step

  mysql -u root -p

CREATE DATABASE opendmarc;
CREATE USER 'dmarc'@'localhost' IDENTIFIED BY 'dmarcdatabaseuserpassword';
GRANT ALL PRIVILEGES ON `opendmarc` . * TO 'dmarc'@'localhost';
FLUSH PRIVILEGES;
quit;

once we’ve created database and user, next step is populate tables.

mysql -u dmarc -p opendmarc < /usr/share/doc/opendmarc-1.3.1/schema.mysql

should be easy now we can configure opendmarc

vim /etc/opendmarc.conf

replace line 15 to have these aspect


AuthservID HOSTNAME

now we will configure opendmarc service


systemctl enable opendmarc
systemctl start opendmarc

next step is integrate opendmarc service with postfix

vim /etc/postfix/main.cf

just modify smtpd_milters values


smtpd_milters = inet:127.0.0.1:8891 , inet:127.0.0.1:8893
non_smtpd_milters = $smtpd_milters

now you can restart postfix

systemctl restart postfix.service

some domains can use aname.domain.com instead top-level domain this can be a problem during dmarc verification, Mozilla maintais a database calle Public Suffix list that can be added to our opendmarc server to make it more effective

just execute these commands

/usr/bin/wget –no-check-certificate -q -N -P /etc/opendmarc https://publicsuffix.org/list/effective_tld_names.dat
chown opendmarc:opendmarc /etc/opendmarc/effective_tld_names.dat

we should maintain this database updated we can add a cron job to do this job

just edit crontab

vim /etc/crontab

and append these line


@weekly /usr/bin/wget --no-check-certificate -q -N -P /etc/opendmarc https://publicsuffix.org/list/effective_tld_names.dat #Get latest effective_tld_names for OpenDMARC

Opendmarc needs know where Pubblic Suffix List is located

vim /etc/opendmarc.conf

line 221 should be like this


PublicSuffixList /etc/opendmarc/effective_tld_names.dat

we will restart opendmarc service to apply configuration changes

systemctl restart opendmarc

now we will configure opendmarc to maintain a history file

vim /etc/opendmarc.conf

uncomment line 166


HistoryFile /var/spool/opendmarc/opendmarc.dat

and restart opendmarc again

systemctl restart opendmarc

now we will process this file with a script every hour

vim /etc/cron.hourly/processdmarc.cron

it should have this content.

Remenber put your dmarc password in dbpass variable

and correct domain settings in /usr/sbin/opendmarc-reports command

#!/bin/bash
# Imports data from OpenDMARC's opendmarc.dat file into a local MySQL DB
# and sends DMARC failure reports to domain owners.
# Based on a script from Hamzah Khan (http://blog.hamzahkhan.com/)

set -e

# Database and History File Info
DBHOST='localhost'
DBUSER='dmarc'
DBPASS='yourpassword'
DBNAME='opendmarc'
HISTDIR='/var/spool/opendmarc'
HISTFILE='opendmarc'

# Make sure history file exists
touch ${HISTDIR}/${HISTFILE}.dat

# Move history file temp dir for processing
mv ${HISTDIR}/${HISTFILE}.dat /tmp/${HISTFILE}.$$

# Import temp history file data and send reports
/usr/sbin/opendmarc-import -dbhost=${DBHOST} -dbuser=${DBUSER} -dbpasswd=${DBPASS} -dbname=${DBNAME} -verbose &amp;lt; /tmp/${HISTFILE}.$$
/usr/sbin/opendmarc-reports -dbhost=${DBHOST} -dbuser=${DBUSER} -dbpasswd=${DBPASS} -dbname=${DBNAME} -verbose -interval=86400 -report-email 'postmaster@cheatcodes.com' -report-org 'CheatCodes.com'
/usr/sbin/opendmarc-expire -dbhost=${DBHOST} -dbuser=${DBUSER} -dbpasswd=${DBPASS} -dbname=${DBNAME} -verbose

# Delete temp history file
rm -rf /tmp/${HISTFILE}.$$

finaly we will make this file executable

chmod +x /etc/cron.hourly/processdmarc.cron

Now your postfix has dmarc support and every hour will send necesary reports

 

Step 43: Configuring Roundcube webmail

Smartphones are a inclredible good technology but sometimes we don’t have signal and we need to check our email, using roundcube we provide a webmail infrastructure to our users, where they can read and send emails using a web browser.

In order to configure roundcube we need to create a database, yes another one, using these  commands

    mysql -u root -p

CREATE DATABASE roundcubemail;
CREATE USER 'roundcube'@'localhost' IDENTIFIED BY 'a_secure_password';
GRANT ALL PRIVILEGES ON `roundcubemail` . * TO 'roundcube'@'localhost';
FLUSH PRIVILEGES;
quit;
 

next step is populate tables

 mysql -uroundcube -p roundcube </usr/share/roundcubemail/SQL/mysql.initial.sql

we also need to enable some selinux options for our case

setsebool -P httpd_can_sendmail on

now we need to configure roundcubemail to access database and know were are located the mailservers

we will move example config file to use as template

mv /etc/roundcubemail/config.inc.php.sample /etc/roundcubemail/config.inc.ph

we need to modify some values


$config['db_dsnw'] = 'mysql://roundcube:your_password@localhost/roundcubemail';

$config['default_host'] = 'tls://yourfqdndomain.com';

$config['smtp_server'] = 'tls://yourfqdndomain.com';

$config['imap_auth_type'] = 'login';
$config['smtp_auth_type'] = 'login';
$config['smtp_port'] = 587;

$config['smtp_user'] = '%u';

$config['smtp_pass'] = '%p';

$config['des_key'] = 'PUTHEREA24CHARACTERRANDOMSTRING';

 

next we need to generate php-fpm pool to server roundcubemail config

we will create a dir for store sessions

mkdir /var/lib/php/yourfqdndomain.com

adjust group ownership

chown root:apache /var/lib/php/yourfqdndomain.com

adjust group permissions

chmod 770 /var/lib/php/yourfqdndomain.com

and set a valid selinux labeling

chcon -R -t httpd_var_run_t /var/lib/php/yourfqdndomain.com

now we can create a valid  php-fpm pool

vim /etc/php-fpm.d/yourfqdndomain.com.conf

adapt this content


[yourfqdndomain]
listen = /var/run/php-fpm/yourfqdndomain.socket
listen.allowed_clients = 127.0.0.1
user = apache
group = apache
pm = dynamic
pm.max_children = 50
pm.start_servers = 5
pm.min_spare_servers = 3
pm.max_spare_servers = 35
slowlog = /var/log/php-fpm/yourfqdndomain.slow.log
rlimit_files = 1024
rlimit_core = 0
chdir = /usr/share/roundcubemail
php_flag[display_errors] = off
php_admin_value[error_log] = /var/log/php-fpm/yourfqdndomain-error.log
php_admin_flag[log_errors] = on
php_admin_value[memory_limit] = 128M
php_value[session.save_handler] = files
php_value[session.save_path] = /var/lib/php/yourfqdndomain.com

once edited we can restart php-fpm service to load new configuration

systemctl restart php-fpm.service

check if any error appears reading this command output

 systemctl status php-fpm.service

next step is create a nginx record and certificate

nginx doesn’t have intermediate certificate config option we need to append pulic key with intermidiate certificate in one file executing a command like this.

cat publiccertificate.crt intermediatecertificate.crt > bundle_certificate.crt

Order of cat command is important public certificate will be at the beginning otherwise you will get ssl errors in nginx.

finally we can create our nginx config file

vim /etc/nginx/conf.d/yourfqdndomain.conf

and adapt these template


server {

listen 80;
server_name yourfqdndomain.com;
error_log  /var/log/nginx/yourfqdndomain.error.log warn;
access_log  /var/log/nginx/yourfqdndomain.access.log;

return 301 https://$server_name$request_uri; # enforce https

}

server {

listen          443 ssl;
server_name    yourfqdndomain;

error_log  /var/log/nginx/yourfqdndomain.secure.error.log warn;
access_log  /var/log/nginx/yourfqdndomain.secure.access.log;

root            /usr/share/roundcubemail;
index           index.php;
charset         utf-8;

## SSL settings
ssl_certificate           /etc/pki/tls/certs/yourfqdndomain.bundle.crt;
ssl_certificate_key       /etc/pki/tls/private/yourfqdndomain.key;

ssl_protocols             TLSv1.2;
ssl_ciphers               &amp;quot;EECDH+ECDSA+AESGCM:EECDH+aRSA+AESGCM:!aNULL:!eNULL:!LOW:!3DES:!MD5:!EXP:!PSK:!SRP:!DSS:!RC4&amp;quot;;
ssl_prefer_server_ciphers on;
ssl_session_cache         shared:SSL:10m;
ssl_session_timeout       10m;
ssl_ecdh_curve            secp521r1;

add_header Strict-Transport-Security max-age=31536000;
# add_header X-Frame-Options DENY;

# auth_basic &amp;quot;Restricted area&amp;quot;;
# auth_basic_user_file /etc/nginx/passwd;

location / {
try_files $uri $uri/ index.php;
}

location ~/(plugins/enigma/home|bin|installer) {
      deny all;
      return 403;
}

location ~* \.php$ {
try_files $uri =404;
fastcgi_split_path_info ^(.+\.php)(/.+)$;
include       fastcgi_params;
fastcgi_pass  unix:/var/run/php-fpm/yourfqdndomain.socket;
fastcgi_index index.php;
fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
}

}

now we should restart nginx

systemctl restart nginx.service

now we can open a web browser write our fqdn and check our roundcube webpage

roundcube mail loginlog with and account and check if everything is working, send an email respond create a folder …

Now you have a working webmail to your users 😛 .

Step 44: Protecting against brute force attacks using fail2ban

At this point we are filtering spam and using a huge number of antispam technologies.  Unfortunately your mailserver will be exposed to internet and a lot of automatic tool will try get your passwords using bruteforce attacks.

To mitigate these actions we will use fail2ban, fail2ban reads system logs and block  temporaly a ip making a bruteforce attacks, generally this means that we attackers avoid our server because temporally block makes bruteforce attack imposible

To make this fail2ban read logs and counts invalid logins when a number of invalid logins is detected fail2ban adds a rule to firewalld blocking source ip of invalid logins during a penalty time, typical brute force attacks abandom because they can’t test more

first we need to enable several triggers in fail2ban

modifiy needed lines to sections look like these

vim /etc/fail2ban/jail.conf


...

[roundcube-auth]
enabled = true
port     = http,https
logpath  = /var/log/roundcubemail/errors

....

[postfix]
enabled = true
port     = smtp,465,submission
logpath  = %(postfix_log)s

....

[dovecot]
enabled = true
port    = pop3,pop3s,imap,imaps,submission,465,sieve
logpath = %(dovecot_log)s

...

[sieve]
enabled = true
port   = smtp,465,submission
logpath = %(dovecot_log)s

now we need to enable fail2ban service

systemctl enable fail2ban.service

and start service

 systemctl start fail2ban.service

Step 45: Enable imap quota

Some users will send a lot of attachments , and they need to know how many free space leave in their mailbox, we will enable imap_quota plugin to avoid support call about that they can’t receive or send mails with titanic attachments.

Responsive of inform about how many free space left in their mailbox is dovecot, we need to enable some plugins.

vim /etc/dovecot/conf.d/10-mail.conf

line 208 should be


mail_plugins = $mail_plugins quota

vim /etc/dovecot/conf.d/20-imap.conf

line 56 should be


mail_plugins = $mail_plugins imap_quota

finally

vim  /etc/dovecot/conf.d/90-quota.conf

line 68 should look like these


quota = maildir:User quota

Now we need reboot dovecot service

 systemctl restart dovecot.service

 

congrats now you should have a funcional mailserver running.

 

References:

http://www.campworld.net/thewiki/pmwiki.php/LinuxServersCentOS/Cent6VirtMailServer

http://dokuwiki.nausch.org/doku.php/centos:mail_c7:spam_7

https://qmail.jms1.net/test-auth.shtml

http://en.linuxreviews.org/HOWTO_Stop_spam_using_Postfix

https://hynek.me/articles/hardening-your-web-servers-ssl-ciphers/

https://www.ssllabs.com/downloads/SSL_TLS_Deployment_Best_Practices.pdf

https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/Security_Guide/sec-Hardening_TLS_Configuration.html

https://www.df.eu/de/service/df-faq/cloudserver/anleitungen/spam-und-virenschutz-mit-postfix-debian/

https://z0z0.me/2014/10/26/install-postfix-dovecot-auth-tls-mysql-postfixadmin-spamassassin-and-clamav-on-centos7/

https://www.2realities.com/blog/2014/02/13/secure-ssl-configuration-for-apache-postfix-dovecot/

https://bettercrypto.org/

https://wiki.gentoo.org/wiki/Complete_Virtual_Mail_Server

http://www.gentoo-wiki.info/Amavisd-new

http://www.iredmail.org/forum/topic3800-iredmail-support-localdomainsmaps-in-amavisdnew.html

http://gogs.info/books/debian-mail/chunked/antispam.amavis.html

http://www.nervous.it/2010/03/amavisd-and-per-user-spam-folder/#Howto

https://github.com/hardware/nginx-config/blob/master/virtual-hosts/https/postfixadmin

http://peketamin.hatenablog.jp/entry/2014/06/26/%28ubuntu_14_04%29_postfix_%E3%82%92%E5%A4%96%E9%83%A8%E3%81%8B%E3%82%89%E6%8E%A5%E7%B6%9A%E5%87%BA%E6%9D%A5%E3%82%8B%E3%82%88%E3%81%86%E3%81%AB%E3%81%99%E3%82%8B_%E2%86%92_dovecot_%E4%BD%BF%E3%81%A3

https://community.rackspace.com/products/f/43/t/51

https://bugzilla.redhat.com/show_bug.cgi?id=1225596

http://www.zytrax.com/books/dns/ch9/dmarc.html

http://dmarc.org/draft-dmarc-base-00-01.txt

http://www.stevejenkins.com/blog/2015/03/installing-opendmarc-rpm-via-yum-with-postfix-or-sendmail-for-rhel-centos-fedora/

Centos 7 Backup MariaDB using MySQL-zrm

One common mistake when we configure a server is forgot run backups, No enough time and other excuses can convince you that make a backup of a new server is not a priority, but sometimes a hipervisor fails and recover data from your virtual machine is hard or impossible. Trust me i saw a lot of unrecoverable databases in development environments unrecoverable after a power outage or hardware fail.

There are a big set of possible ways to backup mysql databases, it depends of several factors like database size, number of databases, reliability …

This tutorial will covered a basic backup for a database server with a small number of small – medium size databases.

Suppose that you have a MariaDB, Percona or Mysql service running and secured,

1 step create a mysql backup user.

  • mysql -uroot -p
  • grant select, insert, update, create, drop, reload, shutdown, alter, super, lock tables, replication client on *.* to ‘zrmbackup’@’localhost’ identified by ‘pass123’;
  • flush PRIVILEGES;

verify permissions

  • show grants for ‘zrmbackup’@’localhost’;

 2 step install Mysql-zrm

yum install epel-release

yum install MySQL-zrm.noarch

3 Step create a backup dir

mkdir /var/backup

*trick in my case I mounted a nfs share in this directory.

4 Step configure Mysql-zrm

We need edit mysql-zrm config file

vim /etc/mysql-zrm/mysql-zrm.conf


#
# Template for Zmanda Recovery Manager for MySQL configuration file
#
# Global configuration file is /etc/mysql-zrm/mysql-zrm.conf
# The file should be copied to /etc/mysql-zrm/<backup set name>/mysql-zrm.conf
# if backup set specific modifications are required.
#
# MySQL ZRM configuration file describes the backup configuration for
# a backup set. This file is organized into five sections for convenience
# - Backup parameters,
# - Databases/tables that are part of backup set,
# - MySQL server parameters
# - ZRM parameters.
# - ZRM plugin parameters.
#
# For more information about Zmanda Recovery Manager for MySQL, please
# see mysql-zrm(1) and/or Administration manual at Zmanda Network.
#
#
# Any line starting with '#' are comments and will be ignored

#
# Backup parameters
#

# Backup comment. This is a text string which can be retrieved
# using the mysql-zrm-reporter(1) tool. You can store some notes
# about the backup set.
# This parameter is optional and has no defaults.
#comment=This is a comment

# Backup level. It can be full or incremental
# Use 0 for full and 1 for incremental backups
# This parameter is optional and default value is full backup.
#
backup-level=0

# Backup method
# Values can be "raw" or "logical". Logical backup are backups using
# mysqldump(1) tool
# This parameter is optional and default value is "raw".
#
backup-mode=logical

# Size of LVM snapshot. This parameter is optional and  is required only
# for "raw" backup-mode and if the MySQL database data are stored in
# LVM logical volumes.
# A size suffix of k for kilobyte, m for megabyte, g for gigabyte
# or t for terabyte
#
#lvm-snapshot=10M

# specifies the plugin for snapshot operations
#
#snapshot-plugin="/usr/share/mysql-zrm/plugins/lvm-snapshot.pl"

# Specifies the type of backup
# Values can be "regular" or "quick".
# Quick backup type uses the snapshot itself as the backup
# without copying the data from the snapshot volume
#backup-type=quick

# Directory to which backups are done. All backups are stored under this
# directory.  This parameter is optional and the default
# value is "/var/lib/mysql-zrm"
#
destination=/var/backup/database

# Specifies how long the backup should be retained. The value can be
# specified in days (suffix D), weeks (suffix: W), months (suffix: M) or
# years (suffix Y). 30 days in a month and 365 days in a year are assumed
# This parameter is optional and the default is the backups are retained
# forever.
#
retention-policy=1Y

# This parameter should be set to 1 if MySQL ZRM backups are being on done on a
# MySQL replication slave.
#replication=1

# This parameter should be set to 1 if backups should be compressed. If this
# parameter is set, gzip(1) command is used by default. If different
# compression algorithm should be used, it must be set in "compress-plugin"
# parameter. Default: There is no data compression.
compress=1

# This specifies the program to be used for compression. The "compression"
# parameter must be set for this parameter to be used. The compression
# command should also support -d option for uncompress backup images. If
# value is not specified then gzip(1) is used for compression.
#compress-plugin=/usr/bin/gzip

# This parameter should be set to 1 if backups should be encrypted.
# The "encrypt-plugin" parameter must be configured. Default: There is no
# data encryption.
#encrypt=1

# This parameter specifies that the program that should be used for
# backup data encryption. "decrypt-option" parameter should also be specified.
#encrypt-plugin="/usr/share/mysql-zrm/plugins/encrypt.pl"

# This specifies the option to be passed to the encryption
# program specified as "encrypt-plugin" parameter for decryption.
#decrypt-option="-d"

#
# Databases/Tables in the backup set
#
# One of the "all-databases" or "databases" or "tables"/"database" parameters
# should be specified. If none of the them are specified, "all-databases"
# is assumed.
#

# This parameter should be set to 1 if all databases are part of this backup set
#
all-databases=1

# List of databases that are part of this backup set. Multiple database
# names are separated by space character. This parameter is ignored if
# "all-databases" is set 1.
#
#databases=wikidb forums

# List of specific tables that are part of this backup set. This parameter
# should not be specified if all tables in the databases in "databases"
# parameter are part of the backup set. Multiple table names should be
# separated by space character. The database to which these tables belong
# to should be specified in "database" parameter.
#
#tables=text user page
#database="wikidb"

#
# The list of databases or tables that are excluded from the backup if the
# database name or table name matches the pattern. Wildcard characters *, ?,
# [, ] are supported. See mysql-zrm-backup man page for details
#
# exclude-pattern=<pattern>

# MySQL server parameters
#

# MySQL database user used for backup and recovery of the backup set.
# This parameter is optional. If this parameter is not specified, values from
# my.cnf configuration file.
#
user="zrmbackup"

# MySQL database user password.
# This parameter is optional. If this parameter is not specified, values from
# my.cnf configuration file or no password is used.
#
password="pass123"

# Fully qualified domain name of the MySQL server.
# This parameter is optional. If this parameter is not specified, values from
# my.cnf configuration file.
#
#host="localhost.company.com"

# Port to which the MySQL server is listening to. This parameter is optional
# and default value is 3306
#
#port=3306

#Name of Socket file that can be used for connecting to MySQL
#
#socket=/var/lib/mysql/mysql.sock

# ssl-options are arguments that are passed to MySQL client commands
# for SSL connection to the MySQL server. This parameter is optional and is
# required only if MySQL server allows SSL connections.
#
#ssl-options="--ssl --ssl-ca=file1 --ssl-cert=file2 --ssl-key=file3"

# This can be set to specify that mysqldump should dump stored routines also.
# This paramter is optional and the default is that stored routines are
# not dumped my mysqldump
routines=1

# This can be set to 0 to specify that the --single-transaction
# should not be used for mysqldump
single-transaction=1

# This can be used to specif the character set name that mysqldump should
# use as default. This parameter is optional.
# If not specified utf8 is used as the default character set.
#default-character-set=latin1

# Directory where MySQL commands can be found. The parameter is optional.
#
#mysql-binpath="/opt/lampp/bin"

# Directory where MySQL binary logs can be found. The parameter is optional.
#
#mysql-binlog-path="/var/log/mysql"

# Directory to use for temporary storage. This parameter is optional
#
#tmpdir=/tmp

#
# ZRM parameters
#

# This parameter controls the verbosity of MySQL ZRM logging. The MySQL ZRM logs
# are available at /var/log/mysql-zrm/mysql-zrm.log. This parameter is optional
# default value is 0 (less verbose).
# The valid values are 0 and 1
#
verbose=1

# After a backup run the backup report is emailed to the mailto address
# This parameter is optional and default behavior is not to send mail
# notifications.
#
mailto="backups@youremail.com"

# Policy on when the mail should be sent
# Values can be "always", "never" or "only-on-error"
mail-policy=only-on-error

# The list of backup reports that are generated after each backup run if
# "html-report-directory" parameter is specified.
# If this parameter is not specified, "backup-status-info" report is generated.
# Valid report names are : backup-method-info, backup-status-info,
#                          backup-retention-info, backup-performance-info,
#                          restore-full-info, restore-incr-info,
#                          replication-info, backup-app-performance-info
# See mysql-zrm-reporter(1) for details of backup reports.
# Multiple report names should be separated by ",".
#
html-reports=backup-status-info

# Directory in which Text/HTML reports will be created by mysql-zrm-reporter(1)
# tool. If this parameter is specified, the mysql-zrm-reporter(1) creates the
# backup reports in this directory after each successful or unsuccessful
# backup run.
# Text reports will be created under "Text" sub-directory
# HTML reports will be created under "Html" sub-directory
#
#html-report-directory=/var/www/mysql-zrm/reports/

# If backup reports are required as RSS feed, "webserver-url" parameter must
# be specified. The value must be set to a valid location on the web server
# in which HTML reports are located and that URL can be used by
# administrator/user to browse HTML reports and can get to the RSS feeds.
# If this parameter is not specified, backup reports are not generated as
# RSS feeds. The list of reports that are available as RSS feed is specified
# in "html-reports".
#
#webserver-url=http://www.company.com/reports/html/

# Location of RSS header file. Administrators can customize RSS channel
# properties using this file. A template for RSS header is available in
# /usr/share/mysql-zrm/plugins/RSS.header file. Location of RSS header
# must be provided if "webserver-url" is specified.
#
#rss-header-location=/etc/mysql-zrm/

#
# ZRM plugin parameters.
# ZRM provides plugin interfaces to allow MySQL administrators to customize
# the backup to their environment.
#

# COPY plugin: Only one copy-plugin must be configured for a backup set.
#
# Socket Copy plugin is to used to transfer backup files from MySQL server to
# the machine running ZRM for MySQL with sockets.
#
# Please read the Notes at /usr/share/doc/mysql-zrm/README-plugin-socket-copy
#
#copy-plugin=/usr/share/mysql-zrm/plugins/socket-copy.pl

# SSH Copy plugin is to used to transfer backup files from MySQL server to
# the machine running ZRM for MySQL with ssh
#
# Please read the Notes at /usr/share/doc/mysql-zrm/README-plugin-ssh-copy
#
#copy-plugin=/usr/share/mysql-zrm/plugins/ssh-copy.pl

# PRE-BACKUP plugin: Plugin that will be called before a backup run for
# the backup set.
#pre-backup-plugin="/usr/share/mysql-zrm/plugins/pre-backup.pl"

# Set of parameters passed to the pre-backup-plugin. These parameters are
# passed to "pre-backup-plugin" before a backup run for the backup set.
# "pre-backup-plugin" parameter must be specified.
#pre-backup-plugin-options="option1 option2"

# POST-BACKUP plugin: Plugin that will be called after a backup run for
# the backup set.
#post-backup-plugin="/usr/share/mysql-zrm/plugins/post-backup.pl"

# Set of parameters passed to the post-backup-plugin. These parameters are
# passed to "post-backup-plugin" after a backup run for the backup set.
# "post-backup-plugin" parameter must be specified.
#post-backup-plugin-options="option1 option2"

# PRE-SCHEDULER plugin: Plugin that can be used to dynamically determine the
# start time for a backup run.
#pre-scheduler-plugin="/usr/share/mysql-zrm/plugins/pre-scheduler.pl"

# ZRM Plugin configuration parameters

# This parameter is used by the encrypt plugin and
# specifies the file containing the passphrase.
#passfile="/tmp/a.pass"

# This parameter is used by ssh-plugin.pl plugin to specify the user to be
# used to ssh to the  remote host
#ssh-user="root"

# This parameter is used by the ssh-copy.pl and socket-copy.pl plugins
# to specify the location of mysql client binaries on the remote host.
#remote-mysql-binpath="/usr/bin"

# This parameter is used by the socket-copy.pl plugin to specify the port
# to be opened on the remote host.
#socket-remote-port="25300"

# This parameter is used by the windows-copy.pl plugin to specify the port
# to be opened on the windows machine during backup
#windows-backup-port="10080"

# This parameter is used by the windows-copy.pl plugin to specify the port
# to be opened on the windows machine during restore
#windows-restore-port="10081"

Step 5. Launch a backup

mysql-zrm-scheduler -now