![[equation]](paper-1.gif)
SENDMAIL
INSTALLATION AND OPERATION GUIDE
Eric Allman
eric@Sendmail.ORG
Version 8.103
For Sendmail Version 8.8
Sendmail implements a general purpose internetwork mail routing facility under the UNIX® operating system. It is not tied to any one transport protocol -- its function may be likened to a crossbar switch, relaying messages from one domain into another. In the process, it can do a limited amount of message header editing to put the message into a format that is appropriate for the receiving domain. All of this is done under the control of a configuration file.
Due to the requirements of flexibility for sendmail, the configuration file can seem somewhat unapproachable. However, there are only a few basic configurations for most sites, for which standard configuration files have been supplied. Most other configurations can be built by adjusting an existing configuration files incrementally.
Sendmail is based on RFC821 (Simple Mail Transport Protocol), RFC822 (Internet Mail Format Protocol), RFC1123 (Internet Host Requirements), RFC1521 (MIME), RFC1651 (SMTP Service Extensions), RFC1891 (SMTP Delivery Status Notifications), RFC1892 (Multipart/Report), RFC1893 (Mail System Status Codes), RFC1894 (Delivery Status Notifications), and RFC1985 (SMTP Service Extension for Remote Message Queue Starting). However, since sendmail is designed to work in a wider world, in many cases it can be configured to exceed these protocols. These cases are described herein.
Although sendmail is intended to run without the need for monitoring, it has a number of features that may be used to monitor or adjust the operation under unusual circumstances. These features are described.
Section one describes how to do a basic sendmail installation. Section two explains the day-to-day information you should know to maintain your mail system. If you have a relatively normal site, these two sections should contain sufficient information for you to install sendmail and keep it happy. Section three describes some parameters that may be safely tweaked. Section four has information regarding the command line arguments. Section five contains the nitty-gritty information about the configuration file. This section is for masochists and people who must write their own configuration file. Section six describes configuration that can be done at compile time. Section seven gives a brief description of differences in this version of sendmail. The appendixes give a brief but detailed explanation of a number of features not described in the rest of the paper.
WARNING:
Several major changes were introduced in version 8.7.
You should not attempt to use this document
for prior versions of
sendmail.
This page intentionally left blank;
replace it with a blank sheet for double-sided output.
There are two basic steps to installing sendmail. The hard part is to build the configuration table. This is a file that sendmail reads when it starts up that describes the mailers it knows about, how to parse addresses, how to rewrite the message header, and the settings of various options. Although the configuration table is quite complex, a configuration can usually be built by adjusting an existing off-the-shelf configuration. The second part is actually doing the installation, i.e., creating the necessary files, etc.
The remainder of this section will describe the installation of sendmail assuming you can use one of the existing configurations and that the standard installation parameters are acceptable. All pathnames and examples are given from the root of the sendmail subtree, normally /usr/src/usr.sbin/sendmail on 4.4BSD.
If you are loading this off the tape, continue with the next section. If you have a running binary already on your system, you should probably skip to section 1.2.
All sendmail source is in the src subdirectory. If you are running on a 4.4BSD system, compile by typing ``make''. On other systems, you may have to make some other adjustments. On most systems, you can do the appropriate compilation by typing
sh makesendmail
Sendmail supports two different formats for the local (on disk) version of databases, notably the aliases database. At least one of these should be defined if at all possible.
If neither of these are defined, sendmail reads the alias file into memory on every invocation. This can be slow and should be avoided. There are also several methods for remote database access:
Other compilation flags are set in conf.h and should be predefined for you unless you are porting to a new environment.
After making the local system configuration described above, You should be able to compile and install the system. The script ``makesendmail'' is the best approach on most systems:
sh makesendmail
You may be able to install using
sh makesendmail install
Sendmail cannot operate without a configuration file. The configuration defines the mail delivery mechanisms understood at this site, how to access them, how to forward email to remote mail systems, and a number of tuning parameters. This configuration file is detailed in the later portion of this document.
The sendmail configuration can be daunting at first. The world is complex, and the mail configuration reflects that. The distribution includes an m4-based configuration package that hides a lot of the complexity.
These configuration files are simpler than old versions largely because the world has become simpler; in particular, text-based host files are officially eliminated, obviating the need to ``hide'' hosts behind a registered internet gateway.
These files also assume that most of your neighbors use domain-based UUCP addressing; that is, instead of naming hosts as ``host!user'' they will use ``host.domain!user''. The configuration files can be customized to work around this, but it is more complex.
Our configuration files are processed by m4 to facilitate local customization; the directory cf of the sendmail distribution directory contains the source files. This directory contains several subdirectories:
SITE(contessa) SITE(hoptoad) SITE(nkainc) SITE(well)
SITECONFIG(site.config.file, name_of_site, X)
If you are in a new domain (e.g., a company), you will probably want to create a cf/domain file for your domain. This consists primarily of relay definitions: for example, Berkeley's domain definition defines relays for BitNET, CSNET, and UUCP. Of these, only the UUCP relay is particularly specific to Berkeley. All of these are internet-style domain names. Please check to make certain they are reasonable for your domain.
Subdomains at Berkeley are also represented in the cf/domain directory. For example, the domain cs-exposed is the Computer Science subdomain with the local hostname shown to other users; cs-hidden makes users appear to be from the CS.Berkeley.EDU subdomain (with no local host information included). You will probably have to update this directory to be appropriate for your domain.
You will have to use or create .mc files in the cf/cf subdirectory for your hosts. This is detailed in the cf/README file.
This subsection describes the files that comprise the sendmail installation.
The binary for sendmail is located in /usr/sbin[1]. It should be setuid root. For security reasons, /, /usr, and /usr/sbin should be owned by root, mode 755[2].
This is the configuration file for sendmail[3]. This and /etc/sendmail.pid are the only non-library file names compiled into sendmail[4].
The configuration file is normally created using the distribution files described above. If you have a particularly unusual system configuration you may need to create a special version. The format of this file is detailed in later sections of this document.
The newaliases command should just be a link to sendmail:
rm -f /usr/bin/newaliases ln -s /usr/sbin/sendmail /usr/bin/newaliases
The hoststat command should just be a link to sendmail, in a fashion similar to newaliases. This command lists the status of the last mail transaction with all remote hosts. It functions only when the HostStatusDirectory option is set.
This command is also a link to sendmail. It flushes all information that is stored in the HostStatusDirectory tree.
The directory /var/spool/mqueue should be created to hold the mail queue. This directory should be mode 700 and owned by root.
The actual path of this directory is defined in the Q option of the sendmail.cf file.
This is a typical value for the HostStatusDirectory option, containing one file per host that this sendmail has chatted with recently. It is normally a subdirectory of mqueue.
The system aliases are held in ``/etc/aliases''. A sample is given in ``lib/aliases'' which includes some aliases which must be defined:
cp lib/aliases /etc/aliases edit /etc/aliases
Normally sendmail looks at a version of these files maintained by the dbm(3) or db(3) routines. These are stored either in ``/etc/aliases.dir'' and ``/etc/aliases.pag'' or ``/etc/aliases.db'' depending on which database package you are using. These can initially be created as empty files, but they will have to be initialized promptly. These should be mode 644:
cp /dev/null /etc/aliases.dir cp /dev/null /etc/aliases.pag chmod 644 /etc/aliases.* newaliases
It will be necessary to start up the sendmail daemon when your system reboots. This daemon performs two functions: it listens on the SMTP socket for connections (to receive mail from a remote system) and it processes the queue periodically to insure that mail gets delivered when hosts come up.
Add the following lines to ``/etc/rc'' (or ``/etc/rc.local'' as appropriate) in the area where it is starting up the daemons:
if [ -f /usr/sbin/sendmail -a -f /etc/sendmail.cf ]; then (cd /var/spool/mqueue; rm -f [lnx]f*) /usr/sbin/sendmail -bd -q30m & echo -n ' sendmail' >/dev/console fi
Some people use a more complex startup script,
removing zero length qf files and df files for which there is no qf file.
For example, see Figure 1
for an example of a complex startup script.
# remove zero length qf files for qffile in qf* do if [ -r $qffile ] then if [ ! -s $qffile ] then echo -n " <zero: $qffile>" > /dev/console rm -f $qffile fi fi done # rename tf files to be qf if the qf does not exist for tffile in tf* do qffile=`echo $tffile | sed 's/t/q/'` if [ -r $tffile -a ! -f $qffile ] then echo -n " <recovering: $tffile>" > /dev/console mv $tffile $qffile else echo -n " <extra: $tffile>" > /dev/console rm -f $tffile fi done # remove df files with no corresponding qf files for dffile in df* do qffile=`echo $dffile | sed 's/d/q/'` if [ -r $dffile -a ! -f $qffile ] then echo -n " <incomplete: $dffile>" > /dev/console mv $dffile `echo $dffile | sed 's/d/D/'` fi done # announce files that have been saved during disaster recovery for xffile in [A-Z]f* do echo -n " <panic: $xffile>" > /dev/console done Figure 1 -- A complex startup script
If you are not running a version of UNIX that supports Berkeley TCP/IP, do not include the -bd flag.
This is the help file used by the SMTP HELP command. It should be copied from ``lib/sendmail.hf'':
cp lib/sendmail.hf /usr/lib
If you wish to collect statistics about your mail traffic, you should create the file ``/etc/sendmail.st'':
cp /dev/null /etc/sendmail.st chmod 666 /etc/sendmail.st
If sendmail is invoked as ``mailq,'' it will simulate the -bp flag (i.e., sendmail will print the contents of the mail queue; see below). This should be a link to /usr/sbin/sendmail.
The system log is supported by the syslogd(8) program. All messages from sendmail are logged under the facility[5].
Each line in the system log consists of a timestamp, the name of the machine that generated it (for logging from several machines over the local area network), the word ``sendmail:'', and a message[6]. Most messages are a sequence of name=value pairs.
The two most common lines are logged when a message is processed. The first logs the receipt of a message; there will be exactly one of these per message. Some fields may be omitted if they do not contain interesting information. Fields are:
There is also one line logged per delivery attempt (so there can be several per message if delivery is deferred or there are multiple recipients). Fields are:
Not all fields are present in all messages; for example, the relay is not listed for local deliveries.
If you have syslogd(8) or an equivalent installed, you will be able to do logging. There is a large amount of information that can be logged. The log is arranged as a succession of levels. At the lowest level only extremely strange situations are logged. At the highest level, even the most mundane and uninteresting events are recorded for posterity. As a convention, log levels under ten are considered generally ``useful;'' log levels above 64 are reserved for debugging purposes. Levels from 11-64 are reserved for verbose information that some sites might want.
A complete description of the log levels is given in section 4.6.
You can ask sendmail to log a dump of the open files and the connection cache by sending it a signal. The results are logged at priority.
Sometimes a host cannot handle a message immediately. For example, it may be down or overloaded, causing it to refuse connections. The sending host is then expected to save this message in its mail queue and attempt to deliver it later.
Under normal conditions the mail queue will be processed transparently. However, you may find that manual intervention is sometimes necessary. For example, if a major host is down for a period of time the queue may become clogged. Although sendmail ought to recover gracefully when the host comes up, you may find performance unacceptably bad in the meantime.
The contents of the queue can be printed using the mailq command (or by specifying the -bp flag to sendmail):
mailq
Sendmail should run the queue automatically at intervals. The algorithm is to read and sort the queue, and then to attempt to process all jobs in order. When it attempts to run the job, sendmail first checks to see if the job is locked. If so, it ignores the job.
There is no attempt to insure that only one queue processor exists at any time, since there is no guarantee that a job cannot take forever to process (however, sendmail does include heuristics to try to abort jobs that are taking absurd amounts of time; technically, this violates RFC 821, but is blessed by RFC 1123). Due to the locking algorithm, it is impossible for one job to freeze the entire queue. However, an uncooperative recipient host or a program recipient that never returns can accumulate many processes in your system. Unfortunately, there is no completely general way to solve this.
In some cases, you may find that a major host going down for a couple of days may create a prohibitively large queue. This will result in sendmail spending an inordinate amount of time sorting the queue. This situation can be fixed by moving the queue to a temporary place and creating a new queue. The old queue can be run later when the offending host returns to service.
To do this, it is acceptable to move the entire queue directory:
cd /var/spool mv mqueue omqueue; mkdir mqueue; chmod 700 mqueue
To run the old mail queue, run the following command:
/usr/sbin/sendmail -oQ/var/spool/omqueue -q
When the queue is finally emptied, you can remove the directory:
rmdir /var/spool/omqueue
Sendmail stores a large amount of information about each remote system it has connected to in memory. It is now possible to preserve some of this information on disk as well, by using the HostStatusDirectory option, so that it may be shared between several invocations of sendmail. This allows mail to be queued immediately or skipped during a queue run if there has been a recent failure in connecting to a remote machine.
Additionally enabling SingleThreadDelivery has the added effect of single-threading mail delivery to a destination. This can be quite helpful if the remote machine is running an SMTP server that is easily overloaded or cannot accept more than a single connection at a time, but can cause some messages to be punted to a future queue run. It also applies to all hosts, so setting this because you have one machine on site that runs some software that is easily overrun can cause mail to other hosts to be slowed down. If this option is set, you probably want to set the MinQueueAge option as well and run the queue fairly frequently; this will cause hosts that are skipped because another sendmail instance is talking to it to be tried again soon.
The disk based host information is stored in a subdirectory of of the mqueue directory called .hoststat[7]. Removing this directory and its subdirectories has an effect similar to the purgestat command and is completely safe. The information in these directories can be perused with the hoststat command, which will indicate the host name, the last access, and the status of that access. An asterisk in the left most column indicates that a sendmail process currently has the host locked for mail delivery.
The disk based connection information is treated the same way as memory based connection information for the purpose of timeouts. By default, information about host failures is valid for 30 minutes. This can be adjusted with the Timeout.hoststatus option.
The connection information stored on disk may be purged at any time with the purgestat command or by invoking sendmail with the -bH switch. The connection information may be viewed with the hoststat command or by invoking sendmail with the -bh switch.
The implementation of certain system services such as host and user name lookup is controlled by the service switch. If the host operating system supports such a switch sendmail will use the native version. Ultrix, Solaris, and DEC OSF/1 are examples of such systems.
If the underlying operating system does not support a service switch (e.g., SunOS, HP-UX, BSD) then sendmail will provide a stub implementation. The ServiceSwitchFile option points to the name of a file that has the service definitions Each line has the name of a service and the possible implementations of that service. For example, the file:
hosts dns files nis aliases files nis
Service switches are not completely integrated. For example, despite the fact that the host entry listed in the above example specifies to look in NIS, on SunOS this won't happen because the system implementation of gethostbyname(3) doesn't understand this. If there is enough demand sendmail may reimplement gethostbyname(3), gethostbyaddr(3), getpwent(3), and the other system routines that would be necessary to make this work seamlessly.
The alias database exists in two forms. One is a text form, maintained in the file /etc/aliases. The aliases are of the form
name: name1, name2, ...
eric@prep.ai.MIT.EDU: eric@CS.Berkeley.EDU
The second form is processed by the ndbm(3)[9] or db(3) library. This form is in the files /etc/aliases.dir and /etc/aliases.pag. This is the form that sendmail actually uses to resolve aliases. This technique is used to improve performance.
The control of search order is actually set by the service switch. Essentially, the entry
OAswitch:aliases
OA/etc/aliases
aliases nis files nisplus
You can also use alias files. For example, the specification:
OA/etc/aliases OAnis:mail.aliases@my.nis.domain
Additional flags can be added after the colon exactly like a K line -- for example:
OAnis:-N mail.aliases@my.nis.domain
The DB or DBM version of the database may be rebuilt explicitly by executing the command
newaliases
/usr/sbin/sendmail -bi
If the RebuildAliases (old D) option is specified in the configuration, sendmail will rebuild the alias database automatically if possible when it is out of date. Auto-rebuild can be dangerous on heavily loaded machines with large alias files; if it might take more than the rebuild timeout (option AliasWait, old a, which is normally five minutes) to rebuild the database, there is a chance that several processes will start the rebuild process simultaneously.
If you have multiple aliases databases specified, the -bi flag rebuilds all the database types it understands (for example, it can rebuild NDBM databases but not NIS databases).
There are a number of problems that can occur with the alias database. They all result from a sendmail process accessing the DBM version while it is only partially built. This can happen under two circumstances: One process accesses the database while another process is rebuilding it, or the process rebuilding the database dies (due to being killed or a system crash) before completing the rebuild.
Sendmail has three techniques to try to relieve these problems. First, it ignores interrupts while rebuilding the database; this avoids the problem of someone aborting the process leaving a partially rebuilt database. Second, it locks the database source file during the rebuild -- but that may not work over NFS or if the file is unwritable. Third, at the end of the rebuild it adds an alias of the form
@: @
If an error occurs on sending to a certain address, say ``x'', sendmail will look for an alias of the form ``owner-x'' to receive the errors. This is typically useful for a mailing list where the submitter of the list has no control over the maintenance of the list itself; in this case the list maintainer would be the owner of the list. For example:
unix-wizards: eric@ucbarpa, wnj@monet, nosuchuser, sam@matisse owner-unix-wizards: unix-wizards-request unix-wizards-request: eric@ucbarpa
List owners also cause the envelope sender address to be modified. The contents of the owner alias are used if they point to a single user, otherwise the name of the alias itself is used. For this reason, and to obey Internet conventions, the ``owner-'' address normally points at the ``-request'' address; this causes messages to go out with the typical Internet convention of using ``list-request'' as the return address.
If you have a version of sendmail with the user information database compiled in, and you have specified one or more databases using the U option, the databases will be searched for a user:maildrop entry. If found, the mail will be sent to the specified address.
As an alternative to the alias database, any user may put a file with the name ``.forward'' in his or her home directory. If this file exists, sendmail redirects mail for that user to the list of addresses listed in the .forward file. For example, if the home directory for user ``mckusick'' has a .forward file with contents:
mckusick@ernie kirk@calder
Actually, the configuration file defines a sequence of filenames to check. By default, this is the user's .forward file, but can be defined to be more generally using the J option. If you change this, you will have to inform your user base of the change; .forward is pretty well incorporated into the collective subconscious.
Several header lines have special interpretations defined by the configuration file. Others have interpretations built into sendmail that cannot be changed without changing the code. These builtins are described here.
If errors occur anywhere during processing, this header will cause error messages to go to the listed addresses. This is intended for mailing lists.
The Errors-To: header was created in the bad old days when UUCP didn't understand the distinction between an envelope and a header; this was a hack to provide what should now be passed as the envelope sender address. It should go away. It is only used if the UseErrorsTo option is set.
The Errors-To: header is official deprecated and will go away in a future release.
RFC 822 requires at least one recipient field (To:, Cc:, or Bcc: line) in every message. If a message comes in with no recipients listed in the message then sendmail will adjust the header based on the ``NoRecipientAction'' option. One of the possible actions is to add an ``Apparently-To:'' header line for any recipients it is aware of. This is not put in as a standard recipient line to warn any recipients that the list is not complete.
The Apparently-To: header is non-standard and is deprecated.
The Precedence: header can be used as a crude control of message priority. It tweaks the sort order in the queue and can be configured to change the message timeout values.
Sendmail supports the IDENT protocol as defined in RFC 1413. Although this enhances identification of the author of an email message by doing a ``call back'' to the originating system to include the owner of a particular TCP connection in the audit trail it is in no sense perfect; a determined forger can easily spoof the IDENT protocol. The following description is excerpted from RFC 1413:
6. Security Considerations
The information returned by this protocol is at most as trustworthy as the host providing it OR the organization operating the host. For example, a PC in an open lab has few if any controls on it to prevent a user from having this protocol return any identifier the user wants. Likewise, if the host has been compromised the information returned may be completely erroneous and misleading.
The Identification Protocol is not intended as an authorization or access control protocol. At best, it provides some additional auditing information with respect to TCP connections. At worst, it can provide misleading, incorrect, or maliciously incorrect information.
The use of the information returned by this protocol for other than auditing is strongly discouraged. Specifically, using Identification Protocol information to make access control decisions - either as the primary method (i.e., no other checks) or as an adjunct to other methods may result in a weakening of normal host security.
An Identification server may reveal information about users, entities, objects or processes which might normally be considered private. An Identification server provides service which is a rough analog of the CallerID services provided by some phone companies and many of the same privacy considerations and arguments that apply to the CallerID service apply to Identification. If you wouldn't run a "finger" server due to privacy considerations you may not want to run this protocol.
In some cases your system may not work properly with IDENT support due to a bug in the TCP/IP implementation. The symptoms will be that for some hosts the SMTP connection will be closed almost immediately. If this is true or if you do not want to use IDENT, you should set the IDENT timeout to zero; this will disable the IDENT protocol.
The complete list of arguments to sendmail is described in detail in Appendix A. Some important arguments are described here.
The amount of time between forking a process to run through the queue is defined by the -q flag. If you run with delivery mode set to i or b this can be relatively large, since it will only be relevant when a host that was down comes back up. If you run in q mode it should be relatively short, since it defines the maximum amount of time that a message may sit in the queue. (See also the MinQueueAge option.)
RFC 1123 section 5.3.1.1 says that this value should be at least 30 minutes (although that probably doesn't make sense if you use ``queue-only'' mode).
If you allow incoming mail over an IPC connection, you should have a daemon running. This should be set by your /etc/rc file using the -bd flag. The -bd flag and the -q flag may be combined in one call:
/usr/sbin/sendmail -bd -q30m
An alternative approach is to invoke sendmail from inetd(8) (use the -bs flag to ask sendmail to speak SMTP on its standard input and output). This works and allows you to wrap sendmail in a TCP wrapper program, but may be a bit slower since the configuration file has to be re-read on every message that comes in. If you do this, you still need to have a sendmail running to flush the queue:
/usr/sbin/sendmail -q30m
In some cases you may find that the queue has gotten clogged for some reason. You can force a queue run using the -q flag (with no value). It is entertaining to use the -v flag (verbose) when this is done to watch what happens:
/usr/sbin/sendmail -q -v
You can also limit the jobs to those with a particular queue identifier, sender, or recipient using one of the queue modifiers. For example, ``-qRberkeley'' restricts the queue run to jobs that have the string ``berkeley'' somewhere in one of the recipient addresses. Similarly, ``-qSstring'' limits the run to particular senders and ``-qIstring'' limits it to particular queue identifiers.
There are a fairly large number of debug flags built into sendmail. Each debug flag has a number and a level, where higher levels means to print out more information. The convention is that levels greater than nine are ``absurd,'' i.e., they print out so much information that you wouldn't normally want to see them except for debugging that particular piece of code. Debug flags are set using the -d option; the syntax is:
debug-flag: -d debug-list debug-list: debug-option [ , debug-option ]* debug-option: debug-range [ . debug-level ] debug-range: integer | integer - integer debug-level: integer
-d12 Set flag 12 to level 1 -d12.3 Set flag 12 to level 3 -d3-17 Set flags 3 through 17 to level 1 -d3-17.4 Set flags 3 through 17 to level 4
Options can be overridden using the -o or -O command line flags. For example,
/usr/sbin/sendmail -oT2m
/usr/sbin/sendmail -OTimeout.queuereturn=2m
Some options have security implications. Sendmail allows you to set these, but relinquishes its setuid root permissions thereafter[11].
An alternative configuration file can be specified using the -C flag; for example,
/usr/sbin/sendmail -Ctest.cf -oQ/tmp/mqueue
Sendmail gives up its setuid root permissions when you use this flag, so it is common to use a publicly writable directory (such as /tmp) as the spool directory (QueueDirectory or Q option) while testing.
Many SMTP implementations do not fully implement the protocol. For example, some personal computer based SMTPs do not understand continuation lines in reply codes. These can be very hard to trace. If you suspect such a problem, you can set traffic logging using the -X flag. For example,
/usr/sbin/sendmail -X /tmp/traffic -bd
This logs a lot of data very quickly and should NEVER be used during normal operations. After starting up such a daemon, force the errant implementation to send a message to your host. All message traffic in and out of sendmail, including the incoming SMTP traffic, will be logged in this file.
When you build a configuration table, you can do a certain amount of testing using the ``test mode'' of sendmail. For example, you could invoke sendmail as:
sendmail -bt -Ctest.cf
rwset address
3,1,21,4 monet:bollard
If you need more detail, you can also use the ``-d21'' flag to turn on more debugging. For example,
sendmail -bt -d21.99
You should be warned that internally, sendmail applies ruleset 3 to all addresses. In test mode you will have to do that manually. For example, older versions allowed you to use
0 bruce@broadcast.sony.com
3,0 bruce@broadcast.sony.com
As of version 8.7, some other syntaxes are available in test mode: .Dxvalue defines macro x to have the indicated value. This is useful when debugging rules that use the $&x syntax. .Ccvalue adds the indicated value to class c. .Sruleset dumps the contents of the indicated ruleset. -ddebug-spec is equivalent to the command-line flag.
When HostStatusDirectory is enabled, information about the status of hosts is maintained on disk and can thus be shared between different instantiations of sendmail. The status of the last connection with each remote host may be viewed with the command:
sendmail -bh
sendmail -bH
There are a number of configuration parameters you may want to change, depending on the requirements of your site. Most of these are set using an option in the configuration file. For example, the line ``O Timeout.queuereturn=5d'' sets option ``Timeout.queuereturn'' to the value ``5d'' (five days).
Most of these options have appropriate defaults for most sites. However, sites having very high mail loads may find they need to tune them as appropriate for their mail load. In particular, sites experiencing a large number of small messages, many of which are delivered to many recipients, may find that they need to adjust the parameters dealing with queue priorities.
All versions of sendmail prior to 8.7 had single character option names. As of 8.7, options have long (multi-character names). Although old short names are still accepted, most new options do not have short equivalents.
This section only describes the options you are most likely to want to tweak; read section 5 for more details.
All time intervals are set using a scaled syntax. For example, ``10m'' represents ten minutes, whereas ``2h30m'' represents two and a half hours. The full set of scales is:
s seconds m minutes h hours d days w weeks
The argument to the -q flag specifies how often a sub-daemon will run the queue. This is typically set to between fifteen minutes and one hour. RFC 1123 section 5.3.1.1 recommends that this be at least 30 minutes.
Timeouts all have option names ``Timeout.suboption''. The recognized suboptions, their default values, and the minimum values allowed by RFC 1123 section 5.3.2 are:
For compatibility with old configuration files, if no suboption is specified, all the timeouts marked with ** are set to the indicated value.
Many of the RFC 1123 minimum values may well be too short. Sendmail was designed to the RFC 822 protocols, which did not specify read timeouts; hence, versions of sendmail prior to version 8.1 did not guarantee to reply to messages promptly. In particular, a ``RCPT'' command specifying a mailing list will expand and verify the entire list; a large list on a slow system may easily take more than five minutes[13]. I recommend a one hour timeout -- since a communications failure during the RCPT phase is rare, a long timeout is not onerous and may ultimately help reduce network load and duplicated messages.
For example, the lines:
O Timeout.command=25m O Timeout.datablock=3h
After sitting in the queue for a few days, a message will time out. This is to insure that at least the sender is aware of the inability to send a message. The timeout is typically set to five days. It is sometimes considered convenient to also send a warning message if the message is in the queue longer than a few hours (assuming you normally have good connectivity; if your messages normally took several hours to send you wouldn't want to do this because it wouldn't be an unusual event). These timeouts are set using the Timeout.queuereturn and Timeout.queuewarn options in the configuration file (previously both were set using the T option).
Since these options are global, and since you can not know a priori how long another host outside your domain will be down, a five day timeout is recommended. This allows a recipient to fix the problem even if it occurs at the beginning of a long weekend. RFC 1123 section 5.3.1.1 says that this parameter should be ``at least 4-5 days''.
The Timeout.queuewarn value can be piggybacked on the T option by indicating a time after which a warning message should be sent; the two timeouts are separated by a slash. For example, the line
OT5d/4h
By setting the ForkEachJob (Y) option, sendmail will fork before each individual message while running the queue. This will prevent sendmail from consuming large amounts of memory, so it may be useful in memory-poor environments. However, if the ForkEachJob option is not set, sendmail will keep track of hosts that are down during a queue run, which can improve performance dramatically.
If the ForkEachJob option is set, sendmail can not use connection caching.
Every message is assigned a priority when it is first instantiated, consisting of the message size (in bytes) offset by the message class (which is determined from the Precedence: header) times the ``work class factor'' and the number of recipients times the ``work recipient factor.'' The priority is used to order the queue. Higher numbers for the priority mean that the message will be processed later when running the queue.
The message size is included so that large messages are penalized relative to small messages. The message class allows users to send ``high priority'' messages by including a ``Precedence:'' field in their message; the value of this field is looked up in the P lines of the configuration file. Since the number of recipients affects the amount of load a message presents to the system, this is also included into the priority.
The recipient and class factors can be set in the configuration file using the RecipientFactor (y) and ClassFactor (z) options respectively. They default to 30000 (for the recipient factor) and 1800 (for the class factor). The initial priority is:
(Remember, higher values for this parameter actually mean that the job will be treated with lower priority.)
The priority of a job can also be adjusted each time it is processed (that is, each time an attempt is made to deliver it) using the ``work time factor,'' set by the RetryFactor (Z) option. This is added to the priority, so it normally decreases the precedence of the job, on the grounds that jobs that have failed many times will tend to fail again in the future. The RetryFactor option defaults to 90000.
Sendmail can be asked to queue (but not deliver) mail if the system load average gets too high using the QueueLA (x) option. When the load average exceeds the value of the QueueLA option, the delivery mode is set to q (queue only) if the QueueFactor (q) option divided by the difference in the current load average and the QueueLA option plus one exceeds the priority of the message -- that is, the message is queued iff:
![[equation]](paper-2.gif)
The QueueFactor option defaults to 600000, so each point of load average is worth 600000 priority points (as described above).
For drastic cases, the RefuseLA (X) option defines a load average at which sendmail will refuse to accept network connections. Locally generated mail (including incoming UUCP mail) is still accepted.
There are a number of delivery modes that sendmail can operate in, set by the DeliveryMode (d) configuration option. These modes specify how quickly mail will be delivered. Legal modes are:
i deliver interactively (synchronously) b deliver in background (asynchronously) q queue only (don't deliver) d defer delvery attempts (don't deliver)
If you run in mode ``q'' (queue only), ``d'' (defer), or ``b'' (deliver in background) sendmail will not expand aliases and follow .forward files upon initial receipt of the mail. This speeds up the response to RCPT commands. Mode ``i'' cannot be used by the SMTP server.
The level of logging can be set for sendmail. The default using a standard configuration table is level 9. The levels are as follows:
Additionally, values above 64 are reserved for extremely verbose debugging output. No normal site would ever set these.
The modes used for files depend on what functionality you want and the level of security you require.
Sendmail can safely be made setuid to root. At the point where it is about to exec(2) a mailer, it checks to see if the userid is zero; if so, it resets the userid and groupid to a default (set by the u and g options). (This can be overridden by setting the S flag to the mailer for mailers that are trusted and must be called as root.) However, this will cause mail processing to be accounted (using sa(8)) to root rather than to the user sending the mail.
If you don't make sendmail setuid to root, it will still run but you lose a lot of functionality and a lot of privacy, since you'll have to make the queue directory world readable. You could also make sendmail setuid to some pseudo-user (e.g., create a user called ``sendmail'' and make sendmail setuid to that) which will fix the privacy problems but not the functionality issues. Also, this isn't a guarantee of security: for example, root occasionally sends mail, and the daemon often runs as root.
At Berkeley we have the alias database (/etc/aliases*) mode 644. While this is not as flexible as if the database were more 666, it avoids potential security problems with a globally writable database.
The database that sendmail actually used is represented by the two files aliases.dir and aliases.pag (both in /etc) (or aliases.db if you are running with the new Berkeley database primitives). The mode on these files should match the mode on /etc/aliases. If aliases is writable and the DBM files (aliases.dir and aliases.pag) are not, users will be unable to reflect their desired changes through to the actual database. However, if aliases is read-only and the DBM files are writable, a slightly sophisticated user can arrange to steal mail anyway.
If your DBM files are not writable by the world or you do not have auto-rebuild enabled (with the AutoRebuildAliases option), then you must be careful to reconstruct the alias database each time you change the text version:
newaliases
When processing the queue, sendmail will try to keep the last few open connections open to avoid startup and shutdown costs. This only applies to IPC connections.
When trying to open a connection the cache is first searched. If an open connection is found, it is probed to see if it is still active by sending a command. It is not an error if this fails; instead, the connection is closed and reopened.
Two parameters control the connection cache. The ConnectionCacheSize (k) option defines the number of simultaneous open connections that will be permitted. If it is set to zero, connections will be closed as quickly as possible. The default is one. This should be set as appropriate for your system size; it will limit the amount of system resources that sendmail will use during queue runs. Never set this higher than 4.
The ConnectionCacheTimeout (K) option specifies the maximum time that any cached connection will be permitted to idle. When the idle time exceeds this value the connection is closed. This number should be small (under ten minutes) to prevent you from grabbing too many resources from other hosts. The default is five minutes.
Control of host address lookups is set by the hosts service entry in your service switch file. If you are on a system that has built-in service switch support (e.g., Ultrix, Solaris, or DEC OSF/1) then your system is probably configured properly already. Otherwise, sendmail will consult the file /etc/service.switch, which should be created. Sendmail only uses two entries: hosts and aliases.
However, some systems (such as SunOS) will do DNS lookups regardless of the setting of the service switch entry. In particular, the system routine gethostbyname(3) is used to look up host names, and many vendor versions try some combination of DNS, NIS, and file lookup in /etc/hosts without consulting a service switch. Sendmail makes no attempt to work around this problem, and the DNS lookup will be done anyway. If you do not have a nameserver configured at all, such as at a UUCP-only site, sendmail will get a ``connection refused'' message when it tries to connect to the name server. If the hosts switch entry has the service ``dns'' listed somewhere in the list, sendmail will interpret this to mean a temporary failure and will queue the mail for later processing; otherwise, it ignores the name server data.
The same technique is used to decide whether to do MX lookups. If you want MX support, you must have ``dns'' listed as a service in the hosts switch entry.
The ResolverOptions (I) option allows you to tweak name server options. The command line takes a series of flags as documented in resolver(3) (with the leading ``RES_'' deleted). Each can be preceded by an optional `+' or `-'. For example, the line
O ResolverOptions=+AAONLY -DNSRCH
Version level 1 configurations turn DNSRCH and DEFNAMES off when doing delivery lookups, but leave them on everywhere else. Version 8 of sendmail ignores them when doing canonification lookups (that is, when using $[ ... $]), and always does the search. If you don't want to do automatic name extension, don't call $[ ... $].
The search rules for $[ ... $] are somewhat different than usual. If the name being looked up has at least one dot, it always tries the unmodified name first. If that fails, it tries the reduced search path, and lastly tries the unmodified name (but only for names without a dot, since names with a dot have already been tried). This allows names such as ``utc.CS'' to match the site in Czechoslovakia rather than the site in your local Computer Science department. It also prefers A and CNAME records over MX records -- that is, if it finds an MX record it makes note of it, but keeps looking. This way, if you have a wildcard MX record matching your domain, it will not assume that all names match.
To completely turn off all name server access on systems without service switch support (such as SunOS) you will have to recompile with -DNAMED_BIND=0 and remove -lresolv from the list of libraries to be searched when linking.
Some sites mount each user's home directory from a local disk on their workstation, so that local access is fast. However, the result is that .forward file lookups are slow. In some cases, mail can even be delivered on machines inappropriately because of a file server being down. The performance can be especially bad if you run the automounter.
The ForwardPath (J) option allows you to set a path of forward files. For example, the config file line
O ForwardPath=/var/forward/$u:$z/.forward.$w
If you create a directory such as /var/forward, it should be mode 1777 (that is, the sticky bit should be set). Users should create the files mode 644.
On systems that have one of the system calls in the statfs(2) family (including statvfs and ustat), you can specify a minimum number of free blocks on the queue filesystem using the MinFreeBlocks (b) option. If there are fewer than the indicated number of blocks free on the filesystem on which the queue is mounted the SMTP server will reject mail with the 452 error code. This invites the SMTP client to try again later.
Beware of setting this option too high; it can cause rejection of email when that mail would be processed without difficulty.
To avoid overflowing your system with a large message, the MaxMessageSize option can be set to set an absolute limit on the size of any one message. This will be advertised in the ESMTP dialogue and checked during message collection.
The PrivacyOptions (p) option allows you to set certain ``privacy'' flags. Actually, many of them don't give you any extra privacy, rather just insisting that client SMTP servers use the HELO command before using certain commands or adding extra headers to indicate possible spoof attempts.
The option takes a series of flag names; the final privacy is the inclusive or of those flags. For example:
O PrivacyOptions=needmailhelo, noexpn
The flags are detailed in section 5.6.
Normally, sendmail deletes the (envelope) sender from any list expansions. For example, if ``matt'' sends to a list that contains ``matt'' as one of the members he won't get a copy of the message. If the -m (me too) command line flag, or if the MeToo (m) option is set in the configuration file, this behaviour is suppressed. Some sites like to run the daemon with -m.
This section describes the configuration file in detail.
There is one point that should be made clear immediately: the syntax of the configuration file is designed to be reasonably easy to parse, since this is done every time sendmail starts up, rather than easy for a human to read or write. On the ``future project'' list is a configuration-file compiler.
The configuration file is organized as a series of lines, each of which begins with a single character defining the semantics for the rest of the line. Lines beginning with a space or a tab are continuation lines (although the semantics are not well defined in many places). Blank lines and lines beginning with a sharp symbol (`#') are comments.
The core of address parsing are the rewriting rules. These are an ordered production system. Sendmail scans through the set of rewriting rules looking for a match on the left hand side (LHS) of the rule. When a rule matches, the address is replaced by the right hand side (RHS) of the rule.
There are several sets of rewriting rules. Some of the rewriting sets are used internally and must have specific semantics. Other rewriting sets do not have specifically assigned semantics, and may be referenced by the mailer definitions or by other rewriting sets.
The syntax of these two commands are:
Macro expansions of the form $x are performed when the configuration file is read. Expansions of the form $&x are performed at run time using a somewhat less general algorithm. This for is intended only for referencing internally defined macros such as $h that are changed at runtime.
The left hand side of rewriting rules contains a pattern. Normal words are simply matched directly. Metasyntax is introduced using a dollar sign. The metasymbols are:
$* Match zero or more tokens $+ Match one or more tokens $- Match exactly one token $=x Match any phrase in class x $~x Match any word not in class x
$-:$+
UCBARPA:eric
$1 UCBARPA $2 eric
Additionally, the LHS can include $@ to match zero tokens. This is not bound to a $n on the RHS, and is normally only used when it stands alone in order to match the null input.
When the left hand side of a rewriting rule matches, the input is deleted and replaced by the right hand side. Tokens are copied directly from the RHS unless they begin with a dollar sign. Metasymbols are:
$n Substitute indefinite token n from LHS $[name$] Canonicalize name $(map key $@arguments $:default $) Generalized keyed mapping function $>n ``Call'' ruleset n $#mailer Resolve to mailer $@host Specify host $:user Specify user
The $n syntax substitutes the corresponding value from a $+, $-, $*, $=, or $~ match on the LHS. It may be used anywhere.
A host name enclosed between $[ and $] is looked up in the host database(s) and replaced by the canonical name[14]. For example, ``$[ftp$]'' might become ``ftp.CS.Berkeley.EDU'' and ``$[[128.32.130.2]$]'' would become ``vangogh.CS.Berkeley.EDU.'' Sendmail recognizes it's numeric IP address without calling the name server and replaces it with it's canonical name.
The $( ... $) syntax is a more general form of lookup; it uses a named map instead of an implicit map. If no lookup is found, the indicated default is inserted; if no default is specified and no lookup matches, the value is left unchanged. The arguments are passed to the map for possible use.
The $>n syntax causes the remainder of the line to be substituted as usual and then passed as the argument to ruleset n. The final value of ruleset n then becomes the substitution for this rule. The $> syntax can only be used at the beginning of the right hand side; it can be only be preceded by $@ or $:.
The $# syntax should only be used in ruleset zero or a subroutine of ruleset zero. It causes evaluation of the ruleset to terminate immediately, and signals to sendmail that the address has completely resolved. The complete syntax is:
$#mailer $@host $:user
Normally, a rule that matches is retried, that is, the rule loops until it fails. A RHS may also be preceded by a $@ or a $: to change this behavior. A $@ prefix causes the ruleset to return with the remainder of the RHS as the value. A $: prefix causes the rule to terminate immediately, but the ruleset to continue; this can be used to avoid continued application of a rule. The prefix is stripped before continuing.
The $@ and $: prefixes may precede a $> spec; for example:
R$+ $: $>7 $1
Substitution occurs in the order described, that is, parameters from the LHS are substituted, hostnames are canonicalized, ``subroutines'' are called, and finally $#, $@, and $: are processed.
There are five rewriting sets
that have specific semantics.
Four of these are related as depicted by figure 1.
Figure 1 -- Rewriting set semantics
D -- sender domain addition
S -- mailer-specific sender rewriting
R -- mailer-specific recipient rewriting
Ruleset three should turn the address into ``canonical form.'' This form should have the basic syntax:
local-part@host-domain-spec
If no ``@'' sign is specified, then the host-domain-spec may be appended (box ``D'' in Figure 1) from the sender address (if the C flag is set in the mailer definition corresponding to the sending mailer).
Ruleset zero is applied after ruleset three to addresses that are going to actually specify recipients. It must resolve to a {mailer, host, user} triple. The mailer must be defined in the mailer definitions from the configuration file. The host is defined into the $h macro for use in the argv expansion of the specified mailer.
Rulesets one and two are applied to all sender and recipient addresses respectively. They are applied before any specification in the mailer definition. They must never resolve.
Ruleset four is applied to all addresses in the message. It is typically used to translate internal to external form.
In addition, ruleset 5 is applied to all local addresses (specifically, those that resolve to a mailer with the `F=5' flag set) that do not have aliases. This allows a last minute hook for local names.
A few extra rulesets are defined as ``hooks'' that can be defined to get special features. They are all named rulesets. The ``check_*'' forms all give accept/reject status; falling off the end or returning normally is an accept, and resolving to $#error is a reject.
The check_relay ruleset is called after a connection is accepted. It is passed
client.host.name $| client.host.address
The check_mail ruleset is passed the user name parameter of the command. It can accept or reject the address.
The check_rcpt ruleset is passed the user name parameter of the command. It can accept or reject the address.
The check_compat ruleset is passed
sender-address $| recipient-address
Some special processing occurs if the ruleset zero resolves to an IPC mailer (that is, a mailer that has ``[IPC]'' listed as the Path in the M configuration line. The host name passed after ``$@'' has MX expansion performed; this looks the name up in DNS to find alternate delivery sites.
The host name can also be provided as a dotted quad in square brackets; for example:
[128.32.149.78]
The host name passed in after the ``$@'' may also be a colon-separated list of hosts. Each is separately MX expanded and the results are concatenated to make (essentially) one long MX list. The intent here is to create ``fake'' MX records that are not published in DNS for private internal networks.
As a final special case, the host name can be passed in as a text string in square brackets:
[ucbvax.berkeley.edu]
Macros are named with a single character or with a word in {braces}. Single character names may be selected from the entire ASCII set, but user-defined macros should be selected from the set of upper case letters only. Lower case letters and special symbols are used internally. Long names beginning with a lower case letter or a punctuation character are reserved for use by sendmail, so user-defined long macro names should begin with an upper case letter.
The syntax for macro definitions is:
Macros are interpolated using the construct $x, where x is the name of the macro to be interpolated. This interpolation is done when the configuration file is read, except in M lines. The special construct $&x can be used in R lines to get deferred interpolation.
Conditionals can be specified using the syntax:
$?x text1 $| text2 $.
Lower case macro names are reserved to have special semantics, used to pass information in or out of sendmail, and special characters are reserved to provide conditionals, etc. Upper case names (that is, $A through $Z) are specifically reserved for configuration file authors.
The following macros are defined and/or used internally by sendmail for interpolation into argv's for mailers or for other contexts. The ones marked ** are information passed into sendmail[16], the ones marked *** are information passed both in and out of sendmail, and the unmarked macros are passed out of sendmail but are not otherwise used internally. These macros are:
Eric Allman <eric@CS.Berkeley.EDU> eric@CS.Berkeley.EDU (Eric Allman)
There are three types of dates that can be used. The $a and $b macros are in RFC 822 format; $a is the time as extracted from the ``Date:'' line of the message (if there was one), and $b is the current date and time (used for postmarks). If no ``Date:'' line is found in the incoming message, $a is set to the current time also. The $d macro is equivalent to the $b macro in UNIX (ctime) format.
The macros $w, $j, and $m are set to the identity of this host. Sendmail tries to find the fully qualified name of the host if at all possible; it does this by calling gethostname(2) to get the current hostname and then passing that to gethostbyname(3) which is supposed to return the canonical version of that host name.[17] Assuming this is successful, $j is set to the fully qualified name and $m is set to the domain part of the name (everything after the first dot). The $w macro is set to the first word (everything before the first dot) if you have a level 5 or higher configuration file; otherwise, it is set to the same value as $j. If the canonification is not successful, it is imperative that the config file set $j to the fully qualified domain name[18].
The $f macro is the id of the sender as originally determined; when mailing to a specific host the $g macro is set to the address of the sender relative to the recipient. For example, if I send to ``bollard@matisse.CS.Berkeley.EDU'' from the machine ``vangogh.CS.Berkeley.EDU'' the $f macro will be ``eric'' and the $g macro will be ``eric@vangogh.CS.Berkeley.EDU.''
The $x macro is set to the full name of the sender. This can be determined in several ways. It can be passed as flag to sendmail. It can be defined in the environment variable. The third choice is the value of the ``Full-Name:'' line in the header if it exists, and the fourth choice is the comment field of a ``From:'' line. If all of these fail, and if the message is being originated locally, the full name is looked up in the /etc/passwd file.
When sending, the $h, $u, and $z macros get set to the host, user, and home directory (if local) of the recipient. The first two are set from the $@ and $: part of the rewriting rules, respectively.
The $p and $t macros are used to create unique strings (e.g., for the ``Message-Id:'' field). The $i macro is set to the queue id on this host; if put into the timestamp line it can be extremely useful for tracking messages. The $v macro is set to be the version number of sendmail; this is normally put in timestamps and has been proven extremely useful for debugging.
The $c field is set to the ``hop count,'' i.e., the number of times this message has been processed. This can be determined by the -h flag on the command line or by counting the timestamps in the message.
The $r and $s fields are set to the protocol used to communicate with sendmail and the sending hostname. They can be set together using the -p command line flag or separately using the -M or -oM flags.
The $_ is set to a validated sender host name. If the sender is running an RFC 1413 compliant IDENT server and the receiver has the IDENT protocol turned on, it will include the user name on that host.
The ${client_name}, ${client_addr}, and ${client_port} macros are set to the name, address, and port number of the SMTP client who is invoking sendmail as a server. These can be used in the check_* rulesets (using the $& deferred evaluation form, of course!).
Classes of phrases may be defined to match on the left hand side of rewriting rules, where a ``phrase'' is a sequence of characters that do not contain space characters. For example a class of all local names for this site might be created so that attempts to send to oneself can be eliminated. These can either be defined directly in the configuration file or read in from another file. Classes are named as a single letter or a word in {braces}. Class names beginning with lower case letters and special characters are reserved for system use. Classes defined in config files may be given names from the set of upper case letters for short names or beginning with an upper case letter for long names.
The syntax is:
CHmonet ucbmonet
CHmonet CHucbmonet
Elements of classes can be accessed in rules using $= or $~. The $~ (match entries not in class) only matches a single word; multi-word entries in the class are ignored in this context.
Some classes have internal meaning to sendmail:
Sendmail can be compiled to allow a scanf(3) string on the F line. This lets you do simplistic parsing of text files. For example, to read all the user names in your system /etc/passwd file into a class, use
FL/etc/passwd %[^:]
Programs and interfaces to mailers are defined in this line. The format is:
Path The pathname of the mailer Flags Special flags for this mailer Sender Rewriting set(s) for sender addresses Recipient Rewriting set(s) for recipient addresses Argv An argument vector to pass to this mailer Eol The end-of-line string for this mailer Maxsize The maximum message length to this mailer Linelimit The maximum line length in the message body Directory The working directory for the mailer Userid The default user and group id to run as Nice The nice(2) increment for the mailer Charset The default character set for 8-bit characters Type The MTS type information (used for error messages)
The following flags may be set in the mailer description. Any other flags may be used freely to conditionally assign headers to messages destined for particular mailers. Flags marked with ** are not interpreted by the sendmail binary; these are the conventionally used to correlate to the flags portion of the H line. Flags marked with *** apply to the mailers for the sender address rather than the usual recipient mailers.
From: usera@hosta To: userb@hostb, userc
From: usera@hosta To: userb@hostb, userc@hosta
Configuration files prior to level 6 assume the `A', `w', `5', `:', `|', `/', and `@' options on the mailer named ``local''.
The mailer with the special name ``error'' can be used to generate a user error. The (optional) host field is an exit status to be returned, and the user field is a message to be printed. The exit status may be numeric or one of the values USAGE, NOUSER, NOHOST, UNAVAILABLE, SOFTWARE, TEMPFAIL, PROTOCOL, or CONFIG to return the corresponding EX_ exit code, or an enhanced error code as described in RFC 1893, Enhanced Mail System Status Codes. For example, the entry:
$#error $@ NOHOST $: Host unknown in this domain
The mailer named ``local'' must be defined in every configuration file. This is used to deliver local mail, and is treated specially in several ways. Additionally, three other mailers named ``prog'', ``*file*'', and ``*include*'' may be defined to tune the delivery of messages to programs, files, and :include: lists respectively. They default to:
Mprog, P=/bin/sh, F=lsD, A=sh -c $u M*file*, P=/dev/null, F=lsDFMPEu, A=FILE M*include*, P=/dev/null, F=su, A=INCLUDE
The Sender and Recipient rewriting sets may either be a simple ruleset id or may be two ids separated by a slash; if so, the first rewriting set is applied to envelope addresses and the second is applied to headers.
The Directory is actually a colon-separated path of directories to try. For example, the definition ``D=$z:/'' first tries to execute in the recipient's home directory; if that is not available, it tries to execute in the root of the filesystem. This is intended to be used only on the ``prog'' mailer, since some shells (such as csh) refuse to execute if they cannot read the home directory. Since the queue directory is not normally readable by unprivileged users csh scripts as recipients can fail.
The Userid specifies the default user and group id to run as, overriding the DefaultUser option (q.v.). If the S mailer flag is also specified, this is the user and group to run as in all circumstances. This may be given as user:group to set both the user and group id; either may be an integer or a symbolic name to be looked up in the passwd and group files respectively. If only a symbolic user name is specified, the group id in the passwd file for that user is used as the group id.
The Charset field is used when converting a message to MIME; this is the character set used in the Content-Type: header. If this is not set, the DefaultCharset option is used, and if that is not set, the value ``unknown-8bit'' is used. WARNING: this field applies to the sender's mailer, not the recipient's mailer. For example, if the envelope sender address lists an address on the local network and the recipient is on an external network, the character set will be set from the Charset= field for the local network mailer, not that of the external network mailer.
The Type= field sets the type information used in MIME error messages as defined by RFC 1894. It is actually three values separated by slashes: the MTA-type (that is, the description of how hosts are named), the address type (the description of e-mail addresses), and the diagnostic type (the description of error diagnostic codes). Each of these must be a registered value or begin with ``X-''. The default is ``dns/rfc822/smtp''.
The format of the header lines that sendmail inserts into the message are defined by the H line. The syntax of this line is:
Some headers have special semantics that will be described later.
There are a number of global options that can be set from a configuration file. Options are represented by full words; some are also representable as single characters for back compatibility. The syntax of this line is:
The options supported (with the old, one character names in brackets) are:
Port Name/number of listening port (defaults to "smtp") Addr Address mask (defaults INADDR_ANY) Family Address family (defaults to INET) Listen Size of listen queue (defaults to 10) SndBufSize Size of TCP send buffer RcvBufSize Size of TCP receive buffer
i Deliver interactively (synchronously) b Deliver in background (asynchronously) q Just queue the message (deliver during queue run) d Defer delivery and all map lookups (deliver during queue run)
<@known1,@known2,@known3:user@unknown>
s Reject undeclared 8-bit data (``strict'') m Convert undeclared 8-bit data to MIME (``mime'') p Pass undeclared 8-bit data (``pass'')
p Print error messages (default) q No messages, just give exit status m Mail back errors w Write back errors (mail if user not logged in) e Mail back errors and give zero exit stat always
public Allow open access needmailhelo Insist on HELO or EHLO command before MAIL needexpnhelo Insist on HELO or EHLO command before EXPN noexpn Disallow EXPN entirely needvrfyhelo Insist on HELO or EHLO command before VRFY novrfy Disallow VRFY entirely restrictmailq Restrict mailq command restrictqrun Restrict -q command line flag noreceipts Don't return success DSNs goaway Disallow essentially all SMTP status queries authwarnings Put X-Authentication-Warning: headers in messages
aliases files hosts dns nis files
initial wait for initial greeting message [5m, 5m] helo reply to HELO or EHLO command [5m, none] mail reply to MAIL command [10m, 5m] rcpt reply to RCPT command [1h, 5m] datainit reply to DATA command [5m, 2m] datablock data block read [1h, 3m] datafinal reply to final ``.'' in data [1h, 10m] rset reply to RSET command [5m, none] quit reply to QUIT command [2m, none] misc reply to NOOP and VERB commands [2m, none] ident IDENT protocol timeout [30s, none] fileopen** timeout on opening .forward and :include: files [60s, none] command** command read [1h, 5m] queuereturn** how long until a message is returned [5d, 5d] queuewarn** how long until a warning is sent [none, none] hoststatus** how long until host status is ``stale'' [30m, none]
All options can be specified on the command line using the -O or -o flag, but most will cause sendmail to relinquish its setuid permissions. The options that will not cause this are MinFreeBlocks [b], DeliveryMode [d], ErrorMode [e], IgnoreDots [i], LogLevel [L], MeToo [m], OldStyleHeaders [o], PrivacyOptions [p], Timeouts [r], SuperSafe [s], Verbose [v], CheckpointInterval [C], and SevenBitInput [7]. Also, M (define macro) when defining the r or s macros is also considered ``safe''.
Values for the ``Precedence:'' field may be defined using the P control line. The syntax of this field is:
Pname=num
Pfirst-class=0 Pspecial-delivery=100 Plist=-30 Pbulk=-60 Pjunk=-100
To provide compatibility with old configuration files, the V line has been added to define some very basic semantics of the configuration file. These are not intended to be long term supports; rather, they describe compatibility features which will probably be removed in future releases.
N.B.: these version levels have nothing to do with the version number on the files. For example, as of this writing version 8 config files (specifically, 8.7) used version level 6 configurations.
``Old'' configuration files are defined as version level one. Version level two files make the following changes:
Version level three files allow # initiated comments on all lines. Exceptions are backslash escaped # marks and the $# syntax.
Version level four configurations are completely equivalent to level three for historical reasons.
Version level five configuration files change the default definition of $w to be just the first component of the hostname.
Version level six configuration files change many of the local processing options (such as aliasing and matching the beginning of the address for `|' characters) to be mailer flags; this allows fine-grained control over the special local processing. Level six configuration files may also use long option names. The ColonOkInAddr option (to allow colons in the local-part of addresses) defaults on for lower numbered configuration files; the configuration file requires some additional intelligence to properly handle the RFC 822 group construct.
The V line may have an optional /vendor to indicate that this configuration file uses modifications specific to a particular vendor[21]. You may use ``/Berkeley'' to emphasize that this configuration file uses the Berkeley dialect of sendmail.
Special maps can be defined using the line:
Kmapname mapclass arguments
Maps are referenced using the syntax:
$( map key $@ arguments $: default $)
The arguments are passed to the map for arbitrary use. Most map classes can interpolate these arguments into their values using the syntax ``%n'' (where n is a digit) to indicate the corresponding argument. Argument ``%0'' indicates the database key. For example, the rule
R$- ! $+ $: $(uucp $1 $@ $2 $: %1 @ %0 . UUCP $)
decvax %1@%0.DEC.COM research %1@%0.ATT.COM
The built in map with both name and class ``host'' is the host name canonicalization lookup. Thus, the syntax:
$(host hostname$)
$[hostname$]
There are many defined classes.
Kmap1 ... Kmap2 ... Kseqmap sequence map1 map2
Kali switch aliases
aliases nis files
"49ers::ubell"
Kdequote dequote ... R$- $: $(dequote $1 $) R$- $+ $: $>3 $1 $2
"|someprogram < input > output"
Most of these accept as arguments the same optional flags and a filename (or a mapname for NIS; the filename is the root of the database path, so that ``.db'' or some other extension appropriate for the database type will be added to get the actual database name). Known flags are:
The dbm map appends the strings ``.pag'' and ``.dir'' to the given filename; the two db-based maps append ``.db''. For example, the map specification
Kuucp dbm -o -N /usr/lib/uucpmap
The program makemap(8) can be used to build any of the three database-oriented maps. It takes the following flags:
The sendmail daemon does not have to be restarted to read the new maps as long as you change them in place; file locking is used so that the maps won't be read while they are being updated.[22]
New classes can be added in the routine setupmaps in file conf.c.
If you have a version of sendmail with the user database package compiled in, the handling of sender and recipient addresses is modified.
The location of this database is controlled with the UserDatabaseSpec option.
The database is a sorted (BTree-based) structure. User records are stored with the key:
user-name:field-name
Field names define both the syntax and semantics of the value. Defined fields include:
As of this writing, only a few of these fields are actually being used by sendmail: maildrop and mailname. A finger program that uses the other fields is planned.
When the rewriting rules submit an address to the local mailer, the user name is passed through the alias file. If no alias is found (or if the alias points back to the same address), the name (with ``:maildrop'' appended) is then used as a key in the user database. If no match occurs (or if the maildrop points at the same address), forwarding is tried.
If the first token of the user name returned by ruleset 0 is an ``@'' sign, the user database lookup is skipped. The intent is that the user database will act as a set of defaults for a cluster (in our case, the Computer Science Division); mail sent to a specific machine should ignore these defaults.
When mail is sent, the name of the sending user is looked up in the database. If that user has a ``mailname'' record, the value of that record is used as their outgoing name. For example, I might have a record:
eric:mailname Eric.Allman@CS.Berkeley.EDU
If a ``maildrop'' is found for the user, but no corresponding ``mailname'' record exists, the record ``:default:mailname'' is consulted. If present, this is the name of a host to override the local host. For example, in our case we would set it to ``CS.Berkeley.EDU''. The effect is that anyone known in the database gets their outgoing mail stamped as ``user@CS.Berkeley.EDU'', but people not listed in the database use the local hostname.
The user database is built from a text file using the makemap utility (in the distribution in the makemap subdirectory). The text file is a series of lines corresponding to userdb records; each line has a key and a value separated by white space. The key is always in the format described above -- for example:
eric:maildrop
makemap btree /etc/userdb.db < /etc/userdb
define(`confUSERDB_SPEC', /etc/userdb.db)
There are some configuration changes that can be made by recompiling sendmail. This section describes what changes can be made and what has to be modified to make them. In most cases this should be unnecessary unless you are porting sendmail to a new environment.
These parameters are intended to describe the compilation environment, not site policy, and should normally be defined in src/Makefile.
There are also several compilation flags to indicate the environment such as ``_AIX3'' and ``_SCO_unix_''. See the READ_ME file for the latest scoop on these flags.
Parameters and compilation options are defined in conf.h. Most of these need not normally be tweaked; common parameters are all in sendmail.cf. However, the sizes of certain primitive vectors, etc., are included in this file. The numbers following the parameters are their default value.
This document is not the best source of information for compilation flags in conf.h -- see src/READ_ME or src/conf.h itself.
A number of other compilation options exist. These specify whether or not specific code should be compiled in. Ones marked with ** are 0/1 valued.
The following options are normally turned on in per-operating-system clauses in conf.h.
The are several built-in ways of computing the load average. Sendmail tries to auto-configure them based on imperfect guesses; you can select one using the cc option -DLA_TYPE=type, where type is:
If type or is specified, you may also need to specify (the path to your system binary) and (the name of the variable containing the load average in the kernel; usually ``_avenrun'' or ``avenrun'').
The following changes can be made in conf.c.
Not all header semantics are defined in the configuration file. Header lines that should only be included by certain mailers (as well as other more obscure semantics) must be specified in the HdrInfo table in conf.c. This table contains the header name (which should be in all lower case) and a set of header control flags (described below), The flags are:
Let's look at a sample HdrInfo specification:
struct hdrinfo HdrInfo[] =
{
/* originator fields, most to least significant */
"resent-sender", H_FROM,
"resent-from", H_FROM,
"sender", H_FROM,
"from", H_FROM,
"full-name", H_ACHECK,
"errors-to", H_FROM|H_ERRORSTO,
/* destination fields */
"to", H_RCPT,
"resent-to", H_RCPT,
"cc", H_RCPT,
"bcc", H_RCPT|H_STRIPVAL,
/* message identification and control */
"message", H_EOH,
"text", H_EOH,
/* trace fields */
"received", H_TRACE|H_FORCE,
/* miscellaneous fields */
"content-transfer-encoding", H_CTE,
"content-type", H_CTYPE,
NULL, 0,
};
There are a number of important points here. First, header fields are not added automatically just because they are in the HdrInfo structure; they must be specified in the configuration file in order to be added to the message. Any header fields mentioned in the configuration file but not mentioned in the HdrInfo structure have default processing performed; that is, they are added unless they were in the message already. Second, the HdrInfo structure only specifies cliched processing; certain headers are processed specially by ad hoc code regardless of the status specified in HdrInfo. For example, the ``Sender:'' and ``From:'' fields are always scanned on ARPANET mail to determine the sender[24]; this is used to perform the ``return to sender'' function. The ``From:'' and ``Full-Name:'' fields are used to determine the full name of the sender if possible; this is stored in the macro $x and used in a number of ways.
If it is necessary to restrict mail through a relay, the checkcompat routine can be modified. This routine is called for every recipient address. It returns an exit status indicating the status of the message. The status accepts the address, queues the message for a later try, and other values (commonly reject the message. It is up to checkcompat to print an error message (using usrerr) if the message is rejected. For example, checkcompat could read:
int
checkcompat(to, e)
register ADDRESS *to;
register ENVELOPE *e;
{
register STAB *s;
s = stab("private", ST_MAILER, ST_FIND);
if (s != NULL && e->e_from.q_mailer != LocalMailer &&
to->q_mailer == s->s_mailer)
{
usrerr("No private net mail allowed through this machine");
return (EX_UNAVAILABLE);
}
if (MsgSize > 50000 && bitnset(M_LOCALMAILER, to->q_mailer))
{
usrerr("Message too large for non-local delivery");
e->e_flags |= EF_NORETURN;
return (EX_UNAVAILABLE);
}
return (EX_OK);
}
The routine getla should return an approximation of the current system load average as an integer. There are several versions included on compilation flags as described above.
New key maps can be added by creating a class initialization function and a lookup function. These are then added to the routine setupmaps.
The initialization function is called as
xxx_map_init(MAP *map, char *mapname, char *args)
The lookup function is called as
xxx_map_lookup(MAP *map, char buf[], int bufsize, char **av, int *statp)
The routine shouldqueue is called to decide if a message should be queued or processed immediately. Typically this compares the message priority to the current load average. The default definition is:
bool
shouldqueue(pri, ctime)
long pri;
time_t ctime;
{
if (CurrentLA < QueueLA)
return (FALSE);
return (pri > (QueueFactor / (CurrentLA - QueueLA + 1)));
}
An implementation wishing to take the actual age of the message into account can also use the ctime parameter, which is the time that the message was first submitted to sendmail. Note that the pri parameter is already weighted by the number of times the message has been tried (although this tends to lower the priority of the message with time); the expectation is that the ctime would be used as an ``escape clause'' to ensure that messages are eventually processed.
The function refuseconnections returns if incoming SMTP connections should be refused. The current implementation is based exclusively on the current load average and the refuse load average option (option X, global variable RefuseLA):
bool
refuseconnections()
{
return (CurrentLA >= RefuseLA);
}
The routine getla returns the current load average (as a rounded integer). The distribution includes several possible implementations. If you are porting to a new environment you may need to add some new tweaks.[25]
The file src/daemon.c contains a number of routines that are dependent on the local networking environment. The version supplied assumes you have BSD style sockets.
In previous releases, we recommended that you modify the routine maphostname if you wanted to generalize $[ ... $] lookups. We now recommend that you create a new keyed map instead.
The following summarizes changes since the last commonly available version of sendmail (5.67). For a detailed list, consult the file RELEASE_NOTES in the root directory of the sendmail distribution.
Instead of closing SMTP connections immediately, those connections are cached for possible future use. The advent of MX records made this effective for mailing lists; in addition, substantial performance improvements can be expected for queue processing.
If two hosts with different names in a single message happen to have the same set of MX hosts, they can be sent in the same transaction. Version 8 notices this and tries to batch the messages.
A number of changes have been made to make sendmail ``conditionally compliant'' (that is, sendmail satisfies all of the ``MUST'' clauses and most but not all of the ``SHOULD'' clauses in RFC 1123).
The major areas of change are (numbers are RFC 1123 section numbers):
The areas in which sendmail is not ``unconditionally compliant'' are:
Version 8 includes both sending and receiving support for Extended SMTP support as defined by RFC 1651 (basic) and RFC 1653 (SIZE); and limited support for RFC 1652 (BODY).
Previous versions of sendmail used the 0200 bit for quoting. This version avoids that use. However, for compatibility with RFC 822, you can set option `7' to get seven bit stripping.
Individual mailers can still produce seven bit output using the `7' mailer flag.
The user database is an as-yet experimental attempt to provide unified large-site name support. We are installing it at Berkeley; future versions may show significant modifications.
The BIND support, particularly for MX records, had a number of annoying ``features'' which have been removed in this release. In particular, these more tightly bind (pun intended) the name server to sendmail, so that the name server resolution rules are incorporated directly into sendmail.
Generalized keyed files is an idea taken directly from sendmail (albeit with a completely different implementation). They can be useful on large sites.
Version 8 also understands YP.
Classes can now be multiple words. For example,
CShofmann.CS.Berkeley.EDU
The $&x construct has been adopted from
The IDENT protocol as defined in RFC 1413 is supported.
A number of small bugs having to do with things like backslash-escaped quotes inside of comments have been fixed.
Since the From: line is passed in separately from the envelope sender, these have both been made visible; the $g macro is set to the envelope sender during processing of mailer argument vectors and the header sender during processing of headers.
It is also possible to specify separate per-mailer envelope and header processing. The SenderRWSet and RecipientRWset arguments for mailers can be specified as envelope/header to give different rewritings for envelope versus header addresses.
When an alias has an associated owner-list name, that alias is used to change the envelope sender address. This will cause downstream errors to be returned to that owner.
The fixed size limit on header lines has been eliminated.
The -B flag has been added to pass in body type information.
The -p flag has been added to pass in protocol information.
The -X flag has been added to allow logging of all protocol in and out of sendmail for debugging.
The -O flag implies setting long-form options.
The -q flag can limit limit a queue run to specific recipients, senders, or queue ids using -qRsubstring, -qSsubstring, or -qIsubstring respectively.
The K line has been added to declare database maps.
The V line has been added to declare the configuration version level.
The M line has a ``D='' field that lets you change into a temporary directory while that mailer is running. It also has a ``U='' field to allow you to set the user and group id to be used when running the mailer.
Several new options have been added, many to support new features, others to allow tuning that was previously available only by recompiling. They are described in detail in Section 5.6. Briefly,
The r (read timeout), I (use BIND), and T (queue timeout) options have been extended to pass in more information.
Several new mailer flags have been added.
All options can be specified using long names, and some new options can only be specified with long names.
The following macros are pre-defined:
Version 8 allows $@ on the Left Hand Side of an ``R'' line to match zero tokens. This is intended to be used to match the null input.
Version 8 allows up to 100 rulesets instead of 30. It is recommended that rulesets 0-9 be reserved for sendmail's dedicated use in future releases.
The total number of MX records that can be used has been raised to 20.
The number of queued messages that can be handled at one time has been raised from 600 to 1000.
Version 8 has changed the default parameters for tuning queue costs to make the number of recipients more important than the size of the message (for small messages). This is reasonable if you are connected with reasonably fast links.
Previously, the ``Full Name <email address>'' syntax would generate incorrect protocol output if ``Full Name'' had special characters such as dot. This version puts quotes around such names.
Several names have been built in to the $@ portion of the $#error mailer.
Previous versions of sendmail treated VRFY and EXPN the same. In this version, VRFY doesn't expand aliases or follow .forward files. EXPN still does.
As an optimization, if you run with your default delivery mode being queue-only or deliver-in-background, the RCPT command will also not chase aliases and .forward files. It will chase them when it processes the queue.
When an address resolves to a mailer that has ``[IPC]'' as its ``Path'', the $@ part (host name) can be a colon-separated list of hosts instead of a single hostname. This asks sendmail to search the list for the first entry that is available exactly as though it were an MX record. The intent is to route internal traffic through internal networks without publishing an MX record to the net. MX expansion is still done on the individual items.
The implementation has been merged with maps. Among other things, this supports NIS-based aliases.
A number of internal changes have been made to enhance portability.
Several fixes have been made to increase the paranoia factor.
Sendmail writes a /etc/sendmail.pid file with the current process id of the SMTP daemon.
Two people using the same program in their .forward file are considered different so that duplicate elimination doesn't delete one of them.
The mailstats program prints mailer names and gets the location of the sendmail.st file from /etc/sendmail.cf.
Many minor bugs have been fixed, such as handling of backslashes inside of quotes.
A hook (ruleset 5) has been added to allow rewriting of local addresses after aliasing.
I've worked on sendmail for many years, and many employers have been remarkably patient about letting me work on a large project that was not part of my official job. This includes time on the INGRES Project at the University of California at Berkeley, at Britton Lee, and again on the Mammoth and Titan Projects at Berkeley.
Much of the second wave of improvements should be credited to Bryan Costales of ICSI. As he passed me drafts of his book on sendmail I was inspired to start working on things again. Bryan was also available to bounce ideas off of.
Many, many people contributed chunks of code and ideas to sendmail. It has proven to be a group network effort. Version 8 in particular was a group project. The following people made notable contributions:
John Beck, Hewlett-Packard Keith Bostic, CSRG, University of California, Berkeley Andrew Cheng, Sun Microsystems Michael J. Corrigan, University of California, San Diego Bryan Costales, International Computer Science Institute Par (Pell) Emanuelsson Craig Everhart, Transarc Corporation Tom Ivar Helbekkmo, Norwegian School of Economics Allan E. Johannesen, WPI Jonathan Kamens, OpenVision Technologies, Inc. Takahiro Kanbe, Fuji Xerox Information Systems Co., Ltd. Brian Kantor, University of California, San Diego Murray S. Kucherawy, HookUp Communication Corp. Bruce Lilly, Sony U.S. Karl London Motonori Nakamura, Ritsumeikan University & Kyoto University John Gardiner Myers, Carnegie Mellon University Neil Rickert, Northern Illinois University Eric Schnoebelen, Convex Computer Corp. Eric Wassenaar, National Institute for Nuclear and High Energy Physics, Amsterdam Christophe Wolfhugel, Pasteur Institute & Herve Schauer Consultants (Paris)
Special thanks are reserved for Michael Corrigan and Christophe Wolfhugel, who besides being wonderful guinea pigs and contributors have also consented to be added to the ``sendmail@Sendmail.ORG'' list and, by answering the bulk of the questions sent to that list, have freed me up to do other work.
Arguments must be presented with flags before addresses. The flags are:
m Deliver mail (default) s Speak SMTP on input side a** ``Arpanet'' mode (get envelope sender information from header) d Run as a daemon in background D Run as a daemon in foreground t Run in test mode v Just verify addresses, don't collect or deliver i Initialize the alias database p Print the mail queue
There are a number of options that may be specified as primitive flags. These are the e, i, m, and v options. Also, the f option may be specified as the -s flag.
This appendix describes the format of the queue files. These files live in the directory defined by the Q option in the sendmail.cf file, usually /var/spool/mqueue or /usr/spool/mqueue.
All queue files have the name xfAAA99999 where AAA99999 is the id for this message and the x is a type. The first letter of the id encodes the hour of the day that the message was received by the system (with A being the hour between midnight and 1:00AM). All files with the same id collectively define one message.
The types are:
The qf file is structured as a series of lines each beginning with a code letter. The lines are as follows:
As an example, the following is a queue file sent to ``eric@mammoth.Berkeley.EDU'' and ``bostic@okeeffe.CS.Berkeley.EDU''[26]:
P835771 T404261372 Seric Ceric:sendmail@vangogh.CS.Berkeley.EDU Reric@mammoth.Berkeley.EDU Rbostic@okeeffe.CS.Berkeley.EDU H?P?return-path: <owner-sendmail@vangogh.CS.Berkeley.EDU> Hreceived: by vangogh.CS.Berkeley.EDU (5.108/2.7) id AAA06703; Fri, 17 Jul 92 00:28:55 -0700 Hreceived: from mail.CS.Berkeley.EDU by vangogh.CS.Berkeley.EDU (5.108/2.7) id AAA06698; Fri, 17 Jul 92 00:28:54 -0700 Hreceived: from [128.32.31.21] by mail.CS.Berkeley.EDU (5.96/2.5) id AA22777; Fri, 17 Jul 92 03:29:14 -0400 Hreceived: by foo.bar.baz.de (5.57/Ultrix3.0-C) id AA22757; Fri, 17 Jul 92 09:31:25 GMT H?F?from: eric@foo.bar.baz.de (Eric Allman) H?x?full-name: Eric Allman Hmessage-id: <9207170931.AA22757@foo.bar.baz.de> HTo: sendmail@vangogh.CS.Berkeley.EDU Hsubject: this is an example message
This is a summary of the support files that sendmail creates or generates. Many of these can be changed by editing the sendmail.cf file; check there to find the actual pathnames.