Software and Hardware Components of Mail Services
This section describes the software and hardware components of a mail system.
Each mail service includes at least one of each of the following software components.
This section also describes these software components.
Mail User Agent
The mail user agent is the program that acts as the interface between the user and
mail transfer agent. The sendmail program is a mail transfer agent. The Solaris operating system
supplies the following mail user agents.
Mail Transfer Agent
The mail transfer agent is responsible for the routing of mail messages and the resolution of
mail addresses. This agent is also known as a mail transport agent. The
transfer agent for the Solaris operating system is sendmail. The transfer agent performs these
Accepts messages from the mail user agent
Resolves destination addresses
Selects a proper delivery agent to deliver the mail
Receives incoming mail from other mail transfer agents
Local Delivery Agent
A local delivery agent is a program that implements a mail delivery protocol. The following local
delivery agents are provided with the Solaris operating system.
The UUCP local delivery agent, which uses uux to deliver mail
The local delivery agent, which is mail.local in the standard Solaris release
Changes From Version 8.12 of sendmail provides information on these related topics.
Mailers and sendmail
Mailer is a sendmail-specific term. A mailer is used by sendmail to identify a specific instance
of a customized local delivery agent or a customized mail transfer agent. You need to
specify at least one mailer in your sendmail.cf file. For task information, refer to
Building the sendmail.cf Configuration File in Chapter 13, Mail Services (Tasks). This section provides a brief description of two types of mailers.
For additional information about mailers, see https://www.sendmail.org/m4/readme.html or /etc/mail/cf/README.
Simple Mail Transfer Protocol (SMTP) Mailers
SMTP is the standard mail protocol that is used on the Internet. This protocol
defines these mailers.
UNIX-to-UNIX Copy Program (UUCP) Mailers
If possible, avoid using UUCP. For an explanation, refer to https://www.sendmail.org/m4/uucp_mailers.html or do
a search in /etc/mail/cf/README on this string: USING UUCP MAILERS.
UUCP defines these mailers.
Names in the $=U class are sent to uucp-old. uucp is the obsolete name for this mailer. The uucp-old mailer uses an exclamation-point address in the headers.
Names in the $=Y class are sent to uucp-new. Use this mailer when you know that the receiving UUCP mailer can manage multiple recipients in one transfer. suucp is the obsolete name for this mailer. The uucp-new mailer also uses an exclamation-point address in the headers.
If MAILER(smtp) is also specified in your configuration, two more mailers are defined.
This mailer uses domain-style addresses and, basically, applies the SMTP rewriting rules.
Names in the $=Z class are sent to uucp-uudom. uucp-uudom and uucp-dom use the same header address format, domain-style addresses.
Note - Because the smtp mailer modifies the UUCP mailer, always put MAILER(smtp) before MAILER(uucp)
in your .mc file.
The mail address contains the name of the recipient and the system to which the
mail message is delivered. When you administer a small mail system that does not
use a name service, addressing mail is easy. The login names uniquely identify the users.
Complexity is introduced if you are administering a mail system that has more than
one system with mailboxes or that has one or more domains. Additional complexity can be
generated if you have a UUCP (or other) mail connection to servers outside your
network. The information in the following sections can help you understand the parts and
complexities of a mail address.
Domains and Subdomains
Email addressing uses domains. A domain is a directory structure for network address naming. A
domain can have one or more subdomains. The domain and subdomains of an
address can be compared to the hierarchy of a file system. Just as a
subdirectory is considered to be inside the directory above it, each subdomain in a
mail address is considered to be inside the location to its right.
The following table shows some top-level domains.
Table 14-7 Top-Level Domains
United States government installations
Other nonprofit organizations
Domains are case insensitive. You can use uppercase, lowercase, or mixed-case letters in the
domain part of an address without making any errors.
Name Service Domain Name and Mail Domain Name
When you are working with name service domain names and mail domain names, remember
By default, the sendmail program strips the first component from the NIS or NIS+ domain name to form the mail domain name. For example, if an NIS+ domain name were bldg5.example.com, its mail domain name would be example.com.
Although mail domain addresses are case insensitive, the NIS or NIS+ domain name is not. For the best results, use lowercase characters when setting up the mail and NIS or NIS+ domain names.
The DNS domain name and the mail domain name must be identical.
For more information, refer to Interactions of sendmail With Name Services.
Typical Format for Mail Addresses
Typically, a mail address has the following format. For further details, refer to Route-Independent Mail Addresses.
[email protected] ... .subdomain2.subdomain1.top-level-domain
The part of the address to the left of the @ sign is the
local address. The local address can contain the following.
Note - The receiving mailer is responsible for determining what the local part of the address
means. For information about mailers, refer to Mailers and sendmail.
The part of the address to the right of the @ sign shows the
domain levels, which is where the local address resides. A dot separates each subdomain.
The domain part of the address can be an organization, a physical area, or a
geographic region. Furthermore, the order of domain information is hierarchical, so the more local
the subdomain, the closer the subdomain is to the @ sign.
Route–Independent Mail Addresses
Mail addresses can be route independent. Route-independent addressing requires the sender of an email
message to specify the name of the recipient and the final destination. A high-speed network,
such as the Internet, uses route-independent addresses. Route-independent addresses can have this format.
Route-independent addresses for UUCP connections can have this address format.
The increased popularity of the domain-hierarchical naming scheme for computers is making route-independent addresses more
common. Actually, the most common route-independent address omits the host name and relies on the
domain name service to properly identify the final destination of the email message.
Route-independent addresses are first read by searching for the @ sign. The domain
hierarchy is then read from the right (the highest level) to the left (the
most specific part of the address to the right of the @ sign).
A mailbox is a file that is the final destination for email messages. The
name of the mailbox can be the user name or the identity of a
specific function, such as the postmaster. Mailboxes are in the /var/mail/username file, which can exist
either on the user's local system or on a remote mail server. In either
instance, the mailbox is on the system to which the mail is delivered.
Mail should always be delivered to a local file system so that the user
agent can pull mail from the mail spool and store it readily in the
local mailbox. Do not use NFS-mounted file systems as the destination for a user's
mailbox. Specifically, do not direct mail to a mail client that is mounting the
/var/mail file system from a remote server. Mail for the user, in this instance,
should be addressed to the mail server and not to the client host name.
NFS-mounted file systems can cause problems with mail delivery and handling.
The /etc/mail/aliases file and name services such as NIS and NIS+ provide mechanisms for
creating aliases for electronic mail addresses. So, users do not need to know the
precise local name of a user's mailbox.
The following table shows some common naming conventions for special-purpose mailboxes.
Table 14-8 Conventions for the Format of Mailbox Names
User names are
frequently the same as mailbox names.
Firstname.Lastname Firstname_Lastname Firstinitial.Lastname Firstinitial_Lastname
can be identified as full names with a dot (or an underscore) that separates
the first and last names. Alternately, user names can be identified by a first
initial with a dot (or an underscore) that separates the initial and the last
Users can address questions and report problems with the mail system to the
postmaster mailbox. Each site and domain should have a postmaster mailbox.
sendmail automatically routes any
mail that is addressed to the MAILER-DAEMON to the postmaster.
Names that end in
-request are administrative addresses for distribution lists. This address should redirect mail to the
person who maintains the distribution list.
Names that begin with owner- are administrative addresses for
distribution lists. This address should redirect mail to the person who handles mail errors.
alias is used when no owner-aliasname alias exists for errors to be returned
to. This address should redirect mail to the person who handles mail errors. This address
also should be defined on any system that maintains a large number of aliases.
The percent sign (%) marks a local address that is
expanded when the message arrives at its destination. Most mail systems interpret mailbox names with
% characters as full mail addresses. The % is replaced with an @, and the
mail is redirected accordingly. Although many people use the % convention, this convention is
not a formal standard. This convention is referred to as the “percent hack.” This feature
is often used to help debug mail problems.
Starting with sendmail version 8, the envelope sender for mail that is sent to
a group alias has been changed to the address that is expanded from the
owner alias, if an owner alias exists. This change enables any mail errors to
be sent to the alias owner, rather than being returned to the sender. With
this change, users notice that mail that was sent to an alias looks as
if the mail came from the alias owner, when delivered. The following alias
format helps with some of the problems that are associated with this change.
mygroup-request: sandys, ignatz
In this example, the mygroup alias is the actual mail alias for the
group. The owner-mygroup alias receives error messages. The mygroup-request alias should be used for
administrative requests. This structure means that in mail sent to the mygroup alias, the envelope
sender changes to mygroup-request.
An alias is an alternate name. For email, you can use aliases to assign
a mailbox location or to define mailing lists. For a task map, refer to
Administering Mail Alias Files (Task Map) in Chapter 13, Mail Services (Tasks). Also, you can refer to Mail Alias Files in this chapter.
For large sites, the mail alias typically defines the location of a mailbox. Providing
a mail alias is like providing a room number as part of the address
for an individual at a large corporation that occupies multiple rooms. If you do
not provide the room number, the mail is delivered to a central address. Without
a room number, extra effort is required to determine where within the building the mail
is to be delivered. So, the possibility of an error increases. For example, if
two people who are named Kevin Smith are in the same building, only one
of them might get mail. To correct the problem, each Kevin Smith should have
a room number added to his address.
Use domains and location-independent addresses as much as possible when you create mailing lists.
To enhance portability and flexibility of alias files, make your alias entries in mailing lists
as generic and system independent as possible. For example, if you have a user
who is named ignatz on system mars, in domain example.com, create the alias [email protected]
instead of [email protected]. If user ignatz changes the name of his system but remains within the
example domain, you do not need to update alias files to reflect the change
in system name.
When you create alias entries, type one alias per line. You should have only
one entry that contains the user's system name. For example, you could create the
following entries for user ignatz.
iggy.ignatz: [email protected]
You can create an alias for local names or domains. For example, an alias
entry for user fred, who has a mailbox on the system mars and
is in the domain planets, could have this entry in the NIS+ aliases
fred: [email protected]
When you create mail lists that include users outside your domain, create the alias
with the user name and the domain name. For example, if you have a
user who is named smallberries on system privet, in domain example.com, create
the alias as [email protected]. The email address of the sender is now automatically translated
to a fully qualified domain name when mail goes outside the user's domain.
The following list describes methods for creating and administering mail alias files.
You can create mail aliases for global use in the NIS+ mail_aliases table, the NIS aliases map, or in local /etc/mail/aliases files. You can also create and administer mailing lists that use the same alias files.
Depending on the configuration of your mail services, you can administer aliases by using the NIS or NIS+ name service to maintain a global aliases database. Otherwise, you could update all the local /etc/mail/aliases files to keep the aliases synchronized.
Users can also create and use aliases. Users can create aliases either in their local ~/.mailrc file, which only the user can use, or in their local /etc/mail/aliases file, which anyone can use. Users cannot normally create or administer NIS or NIS+ alias files.
You can provide the three required elements of mail configuration in the same system
or have separate systems provide these elements.
When users are to communicate with networks outside your domain, you must also add
a fourth element, a mail gateway. For more information, refer to Mail Gateway. The following sections
describe each hardware component.
A mail host is the machine that you designate as the main mail machine on
your network. A mail host is the machine to which other systems at the
site forward mail that cannot be delivered. You designate a system as a mail
host in the hosts database by adding the word mailhost to the right of the
IP address in the local /etc/hosts file. Alternately, you can add the word mailhost
similarly to the hosts file in the name service. For detailed task information, refer
to How to Set Up a Mail Host in Chapter 13, Mail Services (Tasks).
A good candidate for a mail host is a system that is configured as
a router from your network to the Internet global network. For more information, refer
to Chapter 15, Solaris PPP 4.0 (Overview), Chapter 24, UUCP (Overview), and Configuring an IPv4 Router in System Administration Guide: IP Services. If no systems on your local network have a
modem, designate a system as the mail host.
Some sites use standalone machines that are not networked in a time-sharing configuration. Specifically,
the standalone machine serves terminals that are attached to its serial ports. You can set
up electronic mail for this configuration by designating the standalone system as the mail
host of a single-system network. Overview of the Hardware Components in Chapter 12, Mail Services (Overview) provides a figure that shows a
typical email configuration.
A mailbox is a single file that contains email for a particular user. Mail
is delivered to the system where the user's mailbox resides, which can be on
a local machine or a remote server. A mail server is any system that maintains
user mailboxes in its /var/mail directory. For task information, refer to How to Set Up a Mail Server in
Chapter 13, Mail Services (Tasks).
The mail server routes all mail from a client. When a client sends mail,
the mail server puts the mail in a queue for delivery. After the mail
is in the queue, a user can reboot or turn off the client
without losing those mail messages. When the recipient gets mail from a client, the
path in the From line of the message contains the name of the mail server.
If the recipient responds, the response goes to the user's mailbox. Good candidates for mail
servers are systems that provide a home directory for users or systems that are
backed up regularly.
If the mail server is not the user's local system, users in configurations that
use NFS software can mount the /var/mail directory by using the /etc/vfstab file, if they
have root access. Otherwise, users can use the automounter. If NFS support is not
available, users can log in to the server to read their mail.
If users on your network send other types of mail, such as audio files
or files from desktop publishing systems, you need to allocate more space on the
mail server for mailboxes.
By establishing a mail server for all mailboxes, you can simplify your process of
doing backups. Backups can be difficult to do when mail is spread over many
systems. The disadvantage of storing many mailboxes on one server is that the server
can be a single point of failure for many users. However, the advantages of providing
good backups usually make the risk worthwhile.
A mail client is a user of mail services with a mailbox on a
mail server. Additionally, the mail client has a mail alias in the /etc/mail/aliases
file that points to the location of the mailbox. For task information, refer to How to Set Up a Mail Client
in Chapter 13, Mail Services (Tasks).
The mail gateway is a machine that handles connections between networks that run different communications protocols
or communications between different networks that use the same protocol. For example, a mail gateway
might connect a TCP/IP network to a network that runs the Systems Network Architecture
(SNA) protocol suite.
The simplest mail gateway to set up is the mail gateway that connects two
networks that use the same protocol or mailer. This system handles mail with an
address for which sendmail cannot find a recipient in your domain. If a
mail gateway exists, sendmail uses the gateway to send and receive mail outside your domain.
You can set up a mail gateway between two networks that use unmatched mailers,
as shown in the next figure. To support this configuration, you must customize the
sendmail.cf file on the mail gateway system, which can be a difficult and time-consuming
Figure 14-1 Gateway Between Different Communications Protocols
If you have a machine that provides connections to the Internet, you can configure
that machine as the mail gateway. Carefully consider your site's security needs before you
configure a mail gateway. You might need to create a firewall gateway between your corporate
network and other networks, and set up that gateway as the mail gateway. For
task information, refer to How to Set Up a Mail Gateway in Chapter 13, Mail Services (Tasks).