If you want to send or receive an email, you should have a mail server capable of sending and receiving emails across the Internet. In this post, we will discuss Linux mail server and how the SMTP (Simple Mail Transfer Protocol) works as well as other mail-related protocols, like Post Office Protocol (POP) and Internet Message Access Protocol (IMAP) and the relationship between them.
Table of Contents
- 1 Linux SMTP Server
- 2 Mail Service Components
- 3 Setup Email Server
- 4 Linux Mail Server Configuration
- 5 Checking the Mail Queue
- 6 Test Linux Mail Server
- 7 Fight Spam
- 8 Securing SMTP Connection
- 9 POP3 and IMAP Protocol Basics
- 10 Installing Dovecot
- 11 Dovecot Configuration
Linux SMTP Server
SMTP defines how a mail is sent from one host to another, it does not define how the mail should be stored or how it is displayed, and it is also system independent, which means the sender and receiver can have different operating systems.
SMTP requires only that a server is able to send straight ASCII text to another server, and this is done by connecting to the server on port 25 which is the standard SMTP port.
Most Linux distros today are shipped with two of the most common implementations of SMTP which are sendmail and Postfix.
Sendmail is a popular open source mail server implementation used by many Linux distros, but it has a little complex design and less secured.
The Postfix took mail server implementation one step further, it was developed with security in mind.
Mail Service Components
The mail service on any mail server has three components:
Mail user agent (MUA): this component that the user sees and interacts with like Thunderbird and Microsoft Outlook, these user agents are responsible for reading mail and allowing you to compose mail.
Mail transport agent (MTA): this component is responsible for getting the mail from one site to another like Sendmail and Postfix.
Mail delivery agent (MDA): this component is responsible for distributing received messages on the local machine to the appropriate user mailbox like postfix-maildrop and Procmail.
Setup Email Server
We chose Postfix mail server, which is very popular and common between system administrators today.
Postfix is the default mail server on most modern Linux distros.
First, check if it is installed on your system or not:
$ rpm -qa | grep postfix
If not installed, you can install Postfix mail server on Red Hat based distros like this:
$ dnf -y install postfix
Then start the postfix service and enable it on system startup:
$ systemctl start postfix
$ systemctl enable postfix
On Debian based distros like Ubuntu, you can install it like this:
$ apt-get -y install postfix
You will be prompted to select your Postfix mail server configuration type during the installation process.
Among the four choices No configuration, Internet site, Internet with smarthost, Satellite system and Local only, we will choose No configuration option which will create the necessary user and group accounts that Postfix needs.
Linux Mail Server Configuration
After installing the Postfix mail server, you will need to configure it, most of its configuration files can be found under the /etc/postfix/ directory.
You can find the main configuration for Postfix mail server in /etc/postfix/main.cf file.
This file contains a lot of options like:
This option is used for specifying the hostname of the mail server. This is the Internet hostname which Postfix will receive emails on it.
Typical examples of mail server hostnames are mail.example.com, smtp.example.com.
It is written like this:
myhostname = mail.example.com
This option is the mail domain that you will be servicing, like example.com
The syntax is like this:
mydomain = example.com
All emails sent from this mail server will look as though it came from this option. You can set this to $mydomain value.
myorigin = $mydomain
You can use any option value by placing $ in front of the variable name.
This option lists the domains that the Postfix server will take as its final destination for incoming email.
This value is set to the hostname of the server and the domain name, but it can include other names like this:
mydestination = $myhostname, localhost.$mydomain, $mydomain, mail.$mydomain, www.$mydomain
There are two modes of delivery that Postfix mail server can use:
- Directly to a user’s mailbox.
- To a central spool directory, this way, the mail will be in /var/spool/mail with a file for each user.
mail_spool_directory = /var/spool/mail
This variable is an important configuration option. This option allows you to configure what servers can relay through your Postfix server.
You will usually want to allow relaying from local client machines like local mail scripts on your server only.
Otherwise, spammers can utilize your mail server to relay their messages.
If you do not set the mynetworks variable correctly, the spammers may use your mail server as a relay, it is a fast way to get your mail server blacklisted by one of the spam control techniques, such as a DNS Blacklist (DNSBL) or Realtime Blackhole List (RBL). If your server is blacklisted, fewer users will be able to receive emails from other servers.
This option has the following syntax:
mynetworks = 127.0.0.0/8, 192.168.1.0/24
This variable enables you to return a custom response when a client connects to your mail server.
It is better to change the banner to something that doesn’t give an indication about the server you are using.
This variable is used to specify the Internet protocol version that Postfix will use when making a connection.
inet_protocols = ipv4
If you change the configuration files for Postfix mail server, you need to reload the service:
$ systemctl reload postfix
There are more variables in the Postfix configuration file, you may notice that they are commented, these variables allow you to set security levels and debugging levels and other variables.
When you type any configuration, you may make a mistake, you can check for errors using the following command:
$ postfix check
This tool will help you finding exactly the line and the error so you can fix it.
Checking the Mail Queue
Sometimes the mail queues on your system are filled up. This can be caused by many reasons like network failure or any reason that can delay mail delivery.
To check the mail queue on your Linux mail server, use the following command:
This command will display all of the messages that are in the Postfix mail queue.
If your queue is filled up and the message takes several hours to be sent, then you should flush the mail queue.
$ postfix flush
Now, if you check your mail queue you should find it empty.
Test Linux Mail Server
After configuring Postfix mail server correctly, you should test your mail server.
The first step is to use a local mail user agent like mailx or mail which is a symlink to mailx.
Try to send a mail to someone else on the same server, if this works, then send to a remote site.
$ echo "This is message body" | mailx -s "This is Subject" -r "likegeeks<email@example.com>" -a /path/to/attachment firstname.lastname@example.org
Then try to receive a mail from a remote site.
If you have any problems, check the logs. The log file on Red Hat based distros in /var/log/maillog file and on Debian based distros in /var/log/mail.log file or as defined in the rsyslogd configuration.
I recommend you to review the Linux Syslog Server for a detailed explanation about logs and how to configure the rsyslogd.
One of the ways to fight spam is to scan the mailboxes by some tool, searching for certain patterns associated with spam.
One of the best solutions is SpamAssassin, which is open source.
You can install it like this:
$ dnf -y install spamassassin
Then start the service and enable it at startup:
$ systemctl start spamassassin
$ systemctl enable spamassassin
Once you’ve installed it, you can check the configuration in /etc/mail/spamassassin/local.cf file.
SpamAssassin determines if an email is spam or not based on the result of the different scripts scores.
If the message has a higher score, that means a higher possibility of the mail being spam.
In the configuration file, the parameter required_hits 5 indicates that SpamAssassin will mark an email as spam if its score is five or higher.
The report_safe option takes the values 0, 1, or 2. If set to 0 means email marked as spam is sent as it is, only modifying the headers to show that it is spam.
If it takes the value 1 or 2, a new report message is generated by SpamAssassin and sent to the recipient.
If the value is 1, that means the spam message is coded as content message/rfc822, while if the value is 2, that means the message is coded as text/plain content.
The text/plain is safer since some mail clients execute message/rfc822 and could infect the client computer.
Now we need to integrate it into postfix. The simplest way to do this is probably by using procmail.
We’ll have to create a file, named /etc/procmailrc, and add the following content:
Then we edit Postfix configuration file /etc/postfix/main.cf and change mailbox_command like this:
mailbox_command = /usr/bin/procmail
Finally, restart Postfix and SpamAssassin services:
$ systemctl restart postfix
$ systemctl restart spamassassin
However, SpamAssassin sometimes does not recognize spam messages, that led to mailboxes filled with spam messages.
Fortunately, you can filter messages before they enter the Postfix server using Realtime Blackhole Lists (RBLs). That will decrease the load on your mail server and keep your mail server clean.
Open the configuration file of postfix server /etc/postfix/main.cf and change smtpd_recipient_restrictions option and add the following options like this:
strict_rfc821_envelopes = yes
relay_domains_reject_code = 554
unknown_address_reject_code = 554
unknown_client_reject_code = 554
unknown_hostname_reject_code = 554
unknown_local_recipient_reject_code = 554
unknown_relay_recipient_reject_code = 554
unverified_recipient_reject_code = 554
Then restart your postfix server:
$ systemctl restart postfix
The above RBLs are the common ones, you can find more lists on the web and try them.
Securing SMTP Connection
It is better to transfer your SMTP traffic over TLS to protect it from being modified in the middle.
First, we need to generate the certificate and the key using openssl command:
$ openssl genrsa -des3 -out mail.key
$ openssl req -new -key mail.key -out mail.csr
$ cp mail.key mail.key.original
$ openssl rsa -in mail.key.original -out mail_secure.key
$ openssl x509 -req -days 365 -in mail_secure.csr -signkey mail_secure.key -out mail_secure.crt
$ cp mail_secure.crt /etc/postfix/
$ cp mail_secure.key /etc/postfix/
Then add the following option to Postfix configuration file /etc/postfix/main.cf:
smtpd_use_tls = yes
smtpd_tls_cert_file = /etc/postfix/mail_secure.crt
smtpd_tls_key_file = /etc/postfix/mail_secure.key
smtp_tls_security_level = may
Finally, restart your postfix service:
$ systemctl restart postfix
Now, you have to choose the TLS on your client when connecting to the server.
You will receive a warning when you send a mail the first time after changing the setting because of the certificate is not signed.
POP3 and IMAP Protocol Basics
So far we’ve seen how SMTP mail server sends and receives emails without problems, but consider the following situations:
- Users need local copies of e-mail for offline viewing.
- Mail user agents do not support the mbox file format. The mbox format is a simple text format that can be read by a number of console mail user agents like mailx and mutt.
- Users cannot stay connected to a fast network for file system access to their mbox file, so they want to grab a local copy to read offline.
- Security requirements dictate that users not have direct access to the email-gate, for example, shared mail spool directories are considered unacceptable for security reasons.
To handle these cases, another class of protocols was introduced. This type of protocols may be described as mail access protocols.
The most common two popular mail access protocols are Post Office Protocol (POP) and Internet Message Access Protocol (IMAP).
The idea behind POP is very simple: A central Linux mail server remains online all the time and receives and store emails for all users. All received emails are queued on the server until a user connects via POP and downloads the queued mail.
When a user wants to send an email, the email client relays it through the central Linux mail server via SMTP normally.
Note that the SMTP server and POP server can be on the same system without any problem. Most servers do this today.
Features like keeping a master copy of a user’s email on the server with only a cached copy on the client side were missing, that led to the development of IMAP.
By using IMAP, your Linux mail server will support three modes of access:
- The online mode is similar to having direct file system access to the Linux mail server.
- The offline mode is similar to how POP works, where the client is disconnected from the network except when grabbing his email. In this mode, the server normally does not retain a copy of the mail.
- The disconnected mode works by allowing users to keep cached copies of their emails and the server retains a copy of the mail.
There are several implementations for IMAP and POP, the most popular one is Dovecot server which provides both protocols.
The POP3, POP3S, IMAP, and IMAPS listen on ports 110, 995, 143, and 993 respectively.
Most Linux distros come with dovecot preinstalled, however, you can install dovecot in Red Hat based distros like this:
$ dnf -y install dovecot
On Debian based distros, the IMAP and POP3 functionality are provided in two separate packages, you can install them like this:
$ apt-get -y install dovecot-imapd dovecot-pop3d
You will be prompted to create self-signed certificates for using IMAP and POP3 over SSL/TLS. Select yes and enter the hostname for your system when prompted.
Then you can run the service and enable it at startup like this:
$ systemctl start dovecot
$ systemctl enable dovecot
The main configuration file for Dovecot is /etc/dovecot/dovecot.conf file.
Some Linux distros put the configuration under /etc/dovecot/conf.d/ directory and use the include directive to include the settings in the files.
The following list is the some of the parameters used to configure dovecot:
protocols: the protocols you want to support.
protocols = imap pop3 lmtp
lmtp means local mail transfer protocol.
listen: IP addresses to listen on.
listen = *, ::
The asterisk means all ipv4 interfaces and :: means all ipv6 interfaces
userdb: user database for authenticating users.
driver = pam
passdb: password database for authenticating users.
driver = passwd
mail_location: this entry in /etc/dovecot/conf.d/10-mail.conf file, and it is written like this:
mail_location = mbox:~/mail:INBOX=/var/mail/%u
Dovecot comes with generic SSL certificates and key files that are used in the /etc/dovecot/conf.d/10-ssl.conf
ssl_cert = </etc/pki/dovecot/certs/dovecot.pem
ssl_key = </etc/pki/dovecot/private/dovecot.pem
When a user tries to connect to dovecot server, it will show a warning because the certificates are not signed, you can purchase a certificate from a certificate authority if you want.
Don’t forget to open dovecot server ports in your firewall.
$ iptables -A INPUT -p tcp --dport 110 -j ACCEPT
$ iptables -A INPUT -p tcp --dport 995 -j ACCEPT
$ iptables -A INPUT -p tcp --dport 143 -j ACCEPT
$ iptables -A INPUT -p tcp --dport 993 -j ACCEPT
And also SMTP port.
$ iptables -A INPUT -p tcp --dport 25 -j ACCEPT
Then save the rules.
Or if you are using firewalld you can do the following:
$ firewall-cmd --permanent --add-port=110/tcp --add-port=995
$ firewall-cmd --permanent --add-port=143/tcp --add-port=993
$ firewall-cmd --reload
And again, for troubleshooting, you check the log files /var/log/messages, /var/log/maillog, and /var/log/mail.log files.
Linux mail server is one of the easiest servers to work with especially Postfix mail server.
I hope you find the post useful and interesting. Keep coming back.