1 <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
4 <TITLE>EXIM OVERVIEW</TITLE>
5 <META NAME="generator" CONTENT="txt2html v1.21">
7 <BODY bgcolor="#ccccff">
12 Date: 29 November 1996
15 Exim is a mail transport agent (MTA) developed at the University of Cambridge
16 for use on Unix systems connected to the Internet. It is freely available
17 under the terms of the GNU General Public Licence. In style it is similar to
18 Smail 3, but its facilities are more extensive, and in particular it has some
19 defences against mail bombs and unsolicited junk mail, in the form of options
20 for refusing messages from particular hosts, networks, or senders.
23 Exim is in production use on a number of sites that move tens of thousands of
24 messages per day. This document contains an overview description of the way
25 Exim works, with a certain amount of simplification to keep it fairly short.
26 Please address any enquiries about Exim to Philip Hazel:
29 Email: <ph10@cus.cam.ac.uk><BR>
30 Phone: +44 1223 334714<BR>
34 University of Cambridge<BR>
35 Computer Laboratory<BR>
41 This document is copyright (c) University of Cambridge 1996, but copying
42 permission is granted to all.
47 "If I have seen further it is by standing on the shoulders of giants."
56 Exim owes a great deal to Smail 3 and its author, Ron Karr. Without the
57 experience of running and working on the Smail 3 code, I could never have
58 contemplated starting to write a new mailer. Many of the ideas and user
59 interfaces are taken from Smail 3, though the actual code of Exim is entirely
63 My intention was to write a mailer that had more functionality than Smail 3,
64 but which retained the simple lightweight approach, as this seemed to me to be
65 all that was needed for systems directly connected to the Internet, where most
66 messages are delivered almost immediately.
69 <A NAME="0"><H2>2. Availability</H2></A>
72 The current distribution of Exim is available from
75 ftp://ftp.cus.cam.ac.uk/pub/software/programs/exim/exim-n.nn.tar.gz
78 where n.nn is the version number. The distribution contains an ASCII copy of
79 the documentation; other formats are available from
82 ftp://ftp.cus.cam.ac.uk/pub/software/programs/exim/exim-postscript-n.nn.tar.gz
83 ftp://ftp.cus.cam.ac.uk/pub/software/programs/exim/exim-texinfo-n.nn.tar.gz
86 The following operating systems are currently supported: AIX, BSDI, FreeBSD,
87 HP-UX, IRIX, Linux, NetBSD, DEC OSF1 (aka Digital UNIX), SCO, SunOS4, SunOS5,
91 <A NAME="1"><H2>3. Limitations</H2></A>
94 For the benefit of those reading this overview to see whether Exim is of
95 interest to them, its limitations are listed first.
98 <LI> Exim is written in ANSI C. This should not be much of a limitation these
99 days. However, to help with systems that lack a true ANSI C library, Exim
100 avoids making any use of the value returned by the sprintf() function,
101 which is one of the main incompatibilities. It has its own version of
102 strerror() for use with SunOS4 and any other system that lacks this
103 function, and a macro can be defined to turn memmove() into bcopy() if
106 <LI> Exim uses file names that are longer than 14 characters.
108 <LI> Exim is intended for use as an Internet mailer, and therefore handles
109 addresses in RFC 822 domain format only. It cannot handle 'bang paths',
110 though simple two-component bang paths can be converted by a straightforward
111 rewriting configuration.
113 <LI> Exim insists that every address it handles has a domain attached. For
114 incoming local messages, domainless addresses are automatically qualified
115 with a configured domain value. Configuration options specify from which
116 remote systems unqualified addresses are acceptable.
118 <LI> The only external transport currently implemented is an SMTP transport
119 over a TCP/IP network (using sockets), suitable for machines on the
120 Internet. However, a pipe transport is available, and there are facilities
121 for writing messages to files in 'batched SMTP' format; this can be
122 used to send messages to some other transport mechanism. Batched SMTP
123 input is also catered for.
127 <A NAME="2"><H2>4. Main features</H2></A>
130 Exim follows the same general approach of decentralised control that Smail 3
131 does. There is no central process doing overall management of mail delivery.
132 However, unlike Smail, the independent delivery processes share data in the
133 form of 'hints', which makes delivery more efficient in some cases. The hints
134 are kept in a number of DBM files. If any of these files are lost, the only
135 effect is to change the pattern of delivery attempts and retries.
138 Here is a summary of Exim's main features. More details are given in the
139 sections which follow.
142 <LI> Many configuration options can be given as expansion strings, and as
143 these can include file lookups, much of Exim's operation can be made
144 table-driven if desired. For example, it is possible to do local delivery
145 on a machine on which the users do not have accounts.
147 <LI> Regular expressions are available in a number of configuration
150 <LI> Domain lists can include file lookups, making it possible to support a
151 large number of local domains.
153 <LI> Exim has flexible retry algorithms, applicable to mail routing as well as
156 <LI> Exim contains header and envelope rewriting facilities.
158 <LI> Unqualified addresses are accepted only from specified hosts or networks.
160 <LI> Exim can perform multiple deliveries down the same SMTP channel after
161 deliveries to a host have been delayed.
163 <LI> Exim can be configured to do local deliveries immediately but to leave
164 remote deliveries until the message is picked up by a queue-runner
165 process. This increases the likelihood of multiple messages being sent
166 down a single SMTP connection.
168 <LI> When copies of a message have to be delivered to more than one remote
169 host, up to a configured maximum number of remote deliveries can be done
172 <LI> Exim supports optional checking of incoming return path (sender) and
173 receiver addresses as they are received by SMTP.
175 <LI> SMTP calls from specific machines, optionally from specific idents, can
176 be locked out, and incoming SMTP messages from specific senders can also
179 <LI> It is possible to control which hosts may use the Exim host as a relay
180 for onward transmission of mail; the control can be made to depend on the
183 <LI> Messages on the queue can be 'frozen' and 'thawed' by the administrator.
185 <LI> The maximum size of message can be specified.
187 <LI> Exim can handle a number of independent local domains on the same
188 machine; each domain can have its own alias files, etc. These are
189 commonly called "virtual domains".
191 <LI> Exim stats a user's home directory before looking for a .forward file, in
192 order to detect the case of a missing NFS mount.
194 <LI> Exim contains an optional built-in mail filtering facility. This enables
195 users to set up their own mail filtering in a straightforward manner
196 without the need to run an external program. There can also be a system
197 filter file that applies to all messages.
199 <LI> There is support for multiple user mailboxes controlled by prefixes or
200 suffixes on the user name, either via the filter mechanism or through
201 multiple .forward files.
203 <LI> Periodic warnings are automatically sent to messages' senders when
204 delivery is delayed - the time between warnings is configurable.
206 <LI> A queue run can be manually started to deliver just a particular portion
207 of the queue, or those messages with a recipient whose address contains a
210 <LI> Exim can be configured to run as root all the time, except when
211 performing local deliveries, which it always does in a separate process
212 under an appropriate uid and gid. Alternatively, it can be configured to
213 run as root only when needed; in particular, it need not run as root when
214 receiving incoming messages or when sending out messages over SMTP.
216 <LI> I have tried to make the wording of delivery failure messages clearer and
217 simpler, for the benefit of those less-experienced people who are now
220 <LI> The Exim Monitor is an optional extra; it displays information about
221 Exim's processing in an X window, and an administrator can perform a
222 number of control actions from the window interface.
226 <A NAME="3"><H2>5. Performance</H2></A>
229 Although I did not specifically set out to write a high-performance MTA, Exim
230 does seem to be fairly efficient. The busiest site I know of is an ISP that
231 handles over 40,000 messages a day on a Sun Ultra box. Our central mail
232 service machine in Cambridge (a SPARCstation-20) handles over 30,000 messages
233 on a typical day, the volume being around 130 megabytes on the day I looked.
234 The largest number of messages delivered in any one hour was 2753.
237 A system of a different character is sunsite.doc.ic.ac.uk, a SPARCserver 1000
238 system with 8 cpus, which is unusual in that virtually all mail deliveries are
239 remote and relatively large, because it is a data archive that can deliver
240 copies of its holdings via an email interface. On a fairly busy day 14,014
241 messages were received from 231 different hosts and 12,534 deliveries were
242 made to 468 different hosts. The total amount of outgoing mail was 431
243 megabytes. The largest number of deliveries in any one hour was 787.
246 <A NAME="4"><H2>6. Interface</H2></A>
249 Like many MTAs, Exim has adopted the Sendmail interface so that it can be a
250 straight replacement for /usr/lib/sendmail. All the relevant Sendmail options
251 are implemented. There are also some additional options that are compatible
252 with Smail 3, and some further options that are new to Exim.
255 The runtime configuration interface is a single file which is divided into a
256 number of sections. The entries in this file consist of keywords and values,
257 in the style of Smail 3 configuration files.
260 Control of messages on the queue can be done via certain privileged command
261 line options. There is also an optional monitor program called eximon, which
262 displays current information in an X window and contains interfaces to the
263 command line options.
266 <A NAME="5"><H2>7. Method of operation</H2></A>
269 When Exim receives a message, it writes two files in its spool directory. The
270 first contains the envelope information, the current status of the message,
271 and the headers, while the second contains the body of the message. The status
272 of the message includes a complete list of recipients and a list of those that
273 have already received the message. The header file gets updated during the
274 course of delivery if necessary.
277 A message remains in the spool directory until it is completely delivered to
278 its recipients or to an error address, or until it is deleted by an
279 administrator or by the user who originally created it. In cases when delivery
280 cannot proceed - for example, when a message can neither be delivered to its
281 recipients nor returned to its sender, the message is marked 'frozen' on the
282 spool, and no more deliveries are attempted. The administrator can thaw such
283 messages when the problem has been corrected, and can also freeze individual
284 messages by hand if necessary.
287 As delivery proceeds, Exim writes timestamped information about each address
288 to a per-message log file; this includes any delivery error messages. This log
289 is solely for the benefit of the administrator. All the information Exim
290 itself needs for delivery is kept in the header spool file. The message log
291 file is deleted with the spool files. If a message is delayed for more than a
292 configured time, a warning message is sent to the sender. This is repeated
293 whenever the same time elapses again without delivery being complete.
296 The main delivery processing elements of Exim are called directors, routers,
297 and transports. Code for a number of these is provided, and compile-time
298 options specify which ones are actually included in the binary. Directors
299 handle addresses that include one of the local domains, routers handle remote
300 addresses, and transports do actual deliveries.
303 When a message is to be delivered, the sequence of events is roughly as
307 <LI> If there is a system filter file, it is obeyed. This can check on the
308 contents of the message and its headers, and cause delivery to be
309 abandoned or directed to alternative or additional addresses.
311 <LI> Each address is parsed and a check is made to see if it is local or not,
312 by comparing the domain with the list of local domains, which can be
313 wildcarded, or even held in a file if there are a large number of them.
315 <LI> If an address is local, it is passed to each configured director in turn
316 until one is able to handle it. If none can, the address is failed.
317 Directors can be targeted at particular local domains, so several local
318 domains can be processed independently of each other.
320 <LI> A director that accepts an address may set up a local or a remote
321 transport for it, or it may generate one or more new addresses (typically
322 from alias or forward files). New addresses are fed back into this
323 process from the top, but in order to avoid loops, a director will ignore
324 any address which has an identically-named ancestor that was processed by
327 <LI> If an address is not local, it is passed to each router in turn until one
328 is able to handle it. If none can, the address is failed.
330 <LI> A router that accepts an address may set up a transport for it, or may
331 pass an altered address to subsequent routers, or it may discover that
332 the address is a local address after all. This typically happens when an
333 partial domain name is used and (for example) the DNS lookup is
334 configured to try to extend such names. In this case, the address is
335 passed back to the directors.
337 <LI> Routers normally set up remote transports for messages that are to be
338 delivered to other machines. However, a router can pass a message to a
339 local transport, and by this means messages can be routed to other
340 transport mechanisms.
342 <LI> When all the directing and routing is done, addresses that have been
343 successfully handled are passed to their assigned transports. Local
344 transports handle only one address at a time, but remote ones can handle
345 more than one. Each local transport runs in a separate process under a
348 <LI> If there were any errors, a message is returned to an appropriate address
349 (the sender in the common case).
351 <LI> If one or more addresses suffered a temporary failure, the message is
352 left on the queue, to be tried again later. Otherwise the spool files and
353 message log are deleted.
357 <A NAME="6"><H2>8. Mail filtering</H2></A>
360 Exim can be configured to allow users to set up filter files as an alternative
361 to the traditional .forward files. A filter file can test various characteristics
362 of a message, including the contents of the headers and the start of
363 the body, and direct delivery to specified addresses, files, or pipes
364 according to what it finds. The system-wide filter file uses the same control
368 <A NAME="7"><H2>9. Directors</H2></A>
371 The existing directors are listed below. I use the RFC 822 term local-part to
372 mean that portion of an address that comes before the @ character.
375 <LI> aliasfile: This director handles local-part expansion via a traditional
376 alias file. The name of the file is obtained by string expansion, and may
377 therefore depend on the local-part or the domain. Generated pipe and file
378 addresses can be (independently) locked out.
382 The aliasfile director can also be used to test a list of local parts and
383 direct any messages for them to a specific transport. In this case the
384 data associated with the local part in the file is not used for address
385 expansion, but is available for other purposes. For example, files
386 containing records of the form
389 foo: uid=1234 gid=5678 mailbox=/home_1/foo/inbox
392 could be used on a system that did local deliveries without consulting
393 its passwd file. The aliasfile director could use the file to verify that
394 the local part was valid, and then the appendfile transport could use it
395 to get a uid, gid, and mailbox for the delivery.
398 <LI> forwardfile: This director handles local-part expansion via a traditional
399 forward file or, if so configured, by a user's filter file. The name of
400 the file is obtained by string expansion, and may therefore depend on the
401 local-part or the domain, though if it is not an absolute path it is
402 automatically assumed to be in the home directory of the user whose login
403 name is the local-part. Mailing lists can be handled by file names of the
408 /some/list/directory/${local_part}
411 and it is possible to specify an error address for each list that depends
412 on the list name. Generated pipe and file addresses can be (independently)
416 <LI> localuser: This director matches the local-part of an address to a user
417 of the machine. It can also be configured to do a pattern match on the
418 user's home directory name. This makes it possible to partition the set
419 of local users according to their home directories.
421 <LI> smartuser: This director matches any local-part. It can be used to pass
422 messages for unknown users to a script that generates a helpful error
423 message, or it can be used to send such messages to another host,
424 optionally changing the envelope address in the process.
428 The configuration file determines which directors are actually used, and in
429 which order. It is possible to use the same director more than once, with
433 The addresses a director handles can be constrained in the following ways:
436 <LI> A specific set of local domains may be specified, in which case the
437 director is called only for addresses that contain one of those domains.
439 <LI> A specific set of local parts may be specified, in which case the
440 director is called only for addresses that contain one of those local
441 parts. This could be used, for example, to handle 'postmaster' independently
442 of the particular local domain.
444 <LI> A director may be configured to handle local-parts that start with a
445 certain prefix and/or end with a certain suffix. For example, a director
446 can be set up to handle local-parts of the form xxxx-request only.
448 <LI> A flag controls whether a director is called when an address is being
449 verified, as opposed to being directed for delivery.
453 In addition, certain files can be required to exist or not exist for a given
457 <A NAME="8"><H2>10. Routers</H2></A>
460 The existing routers are:
463 <LI> domainlist: This director searches a list of domains for the one it is
464 trying to route. The list may either be a string in the configuration
465 file, possibly including wild cards or regular expressions, or it may be
466 in a file, or both may be provided. In the case of a file, keys of the
467 form *.foo.bar.com can be used for simple wildcarding.
471 If the domain is found, its entry can either specify a single replacement
472 domain name that is passed on to subsequent routers, or it can specify a
473 list of domain names that are looked up by this router. The lookup can be
474 done by the gethostbyname function, or by DNS lookup, and in the latter
475 case it is configurable whether MX or A records or both are used. As well
476 as providing explicit routing for certain domains, the domainlist router
477 can be used to set up gateways for partial domains (e.g. for *.uucp) and
478 it can also be used as a 'smarthost' router by using the all-inclusive
482 <LI> lookuphost: This router looks up domain names either by calling the
483 gethostbyname function, or by using the DNS. In the latter case, it can
484 be configured to use the DNS resolver options for qualifying singlecomponent
485 names and for searching parent domains. It is also possible to
486 specify explicit text strings for widening domains that are not found
487 initially. It is possible to insist on the presence of MX records for
488 certain sets of domains. A configuration option controls whether the
489 message's headers are rewritten when a domain name is changed.
491 <LI> queryprogram: This router passes the address to a script that runs in a
492 separate process under an unprivileged uid and gid. The script returns a
493 line of text specifying whether it matched the domain or not. If it did
494 match, it may specify a transport name, or it may specify that the
495 transport specified for the router is used. The script may also send back
496 a new domain name to replace the current one, and specify a method of
497 looking this name up (gethostbyname, DNS, or pass to next router).
501 The configuration file determines which routers are actually used, and in
502 which order. It is possible to use the same router more than once, with
506 Like directors, routers can be constrained to handle only certain domains or
507 certain local parts (though I haven't seen a good use for that yet). If a
508 router times out, either the delivery can be deferred, or the address can be
509 passed on to the next router.
512 A flag controls whether a router is called when an address is being verified,
513 as opposed to being routed for delivery.
516 <A NAME="9"><H2>11. Transports</H2></A>
519 Local and remote transports are handled differently. A local transport is
520 always run in a separate process with an appropriate real uid and gid. Their
521 values can be specified in the transport's configuration, or passed over from
522 the director that handled the address. The existing transports are:
525 <LI> appendfile: This local transport appends the message to a file whose name
526 is specified as a string containing variable expansions. The current
527 local-part can be inserted via the expansion mechanism, and file names
532 /home/${local_part}/inbox<BR>
533 /var/mail/${local_part}
536 are typical examples. However, it is possible to look up each individual
537 user's inbox name in a file, should that be required.
540 Exclusive access to the file is ensured by using the traditional mailbox
541 locking strategy of creating a lock file. The lock creation process uses
542 a 'hitching post' algorithm (similar to that used by Pine) which is
543 robust when the mailbox file is NFS-mounted. The file is also locked
544 using the lockf function.
547 Options on this transport allow for the insertion of a prefix line (e.g.
548 'From xxx...') and suffix line, special processing of message lines
549 starting with 'From', and the addition of Return-path, Delivery-date, and
550 Envelope-to headers. If the mailbox file is not a regular file, or does
551 not have the correct owner, group, or permissions, no delivery takes
552 place; the address is deferred and the postmaster is informed, except
553 that, if the file's permissions are greater than those required, Exim
554 reduces the permissions and carries on. There are additional checks to
555 reduce the possibility of security exposures caused by race conditions.
558 <LI> pipe: This local transport passes the message via a pipe to a specified
559 command (program or script) which is run in a separate process under a
560 given uid and gid. Various parameters of the message are passed as
561 environment variables, and there are the same options as for appendfile
562 for controlling the form of the message.
566 The returned status of the command may be used to determine success or
567 failure, or it can be ignored. A configuration option specifies whether
568 any standard output generated by the transport is to be returned to the
569 sender. If this is set and output is actually generated, the delivery is
570 deemed to have failed, whatever the returned status of the command. The
571 maximum amount of output generated by the command can be controlled, and
572 a timeout may be set for it.
575 <LI> smtp: This remote transport delivers a message using SMTP over TCP/IP.
576 All addresses in the message that route to the same set of hosts, and
577 have the same errors address (return path), are normally sent in a single
578 transaction. An explicit list of hosts can be set for the transport, or a
579 host list may be attached to an address by one of the routers. If all the
580 hosts are temporarily unable to accept the message, it is delivered to
581 one of a list of fallback hosts, if configured.
585 <A NAME="10"><H2>12. Exim logs</H2></A>
588 Exim write four different log files:
591 <LI> The main log records the arrival of each message and the result of each
592 delivery attempt in a single line in each case. The format is as compact
593 as possible, in an attempt to keep down the size of log files. A number
594 of other events are also recorded on the main log.
596 <LI> The reject log records information from messages that are rejected
597 because their return paths are invalid (a configurable option). The
598 headers are written to this log, following a copy of the one-line message
599 that is also written to the main log. Other types of message rejection
600 also cause writing to this log.
602 <LI> The panic log is written when Exim suffers a disaster and has to bomb
605 <LI> On systems that support signal handlers that restart a system call on
606 exit, Exim reacts to a USR1 signal by writing a line describing its
607 current activity to the process log. This makes it possible to find out
608 what each exim process on a machine is currently doing.
612 A utility script for renaming and compressing the main and reject logs each
613 night is provided. There are also scripts for extracting statistics from log
614 files and for searching log files for the entries for messages that match a
615 given pattern. For example, one can pull out all entries relating to messages
616 for a given local part.
619 <A NAME="11"><H2>13. Exim databases</H2></A>
622 Exim maintains a number of databases in DBM files to help it perform efficient
623 mail delivery. In effect, the files contain hints, and if they are lost it is
624 not a disaster - Exim's performance just suffers a bit. The three databases
628 <LI> retry: This contains information about each failing remote host and
629 temporary failing local delivery - when the first failure was detected,
630 when the delivery (or directing or routing) was last tried, and when it
631 should next be tried. More details about retry algorithms are given
634 <LI> wait-smtp: This contains information about messages that are waiting for
635 particular hosts after an SMTP delivery failure (see the next section).
637 <LI> reject: This contains information about SMTP message rejections (see
642 There is a utility program that lists the contents of one of these databases,
643 and another that allows manual modifications to be applied in some cases.
644 Database records are timestamped, and there is a utility that removes records
645 that are older than a given period, and also cleans up wait-smtp records
646 containing references to messages that no longer exist. Running this daily or
647 weekly should be sufficient to keep the files reasonably tidy.
650 <A NAME="12"><H2>14. SMTP batching</H2></A>
653 When an SMTP delivery attempt fails, causing the message to be deferred till
654 later, Exim updates a DBM database that contains records keyed by host name
655 plus IP address. Each record holds a list of messages that are waiting for
656 that host and address.
659 When an SMTP delivery succeeds, Exim consults the database to see if there are
660 any other messages waiting for the same host and address. If it finds any, it
661 creates a new Exim process and passes it the open SMTP channel and a message
662 identification. The new process then delivers the waiting message down the
663 existing channel and may in turn cause the creation of yet another process.
664 Any other waiting addresses in the message are skipped. The maximum number of
665 messages sent down one connection is configurable.
668 This scheme achieves some SMTP efficiency when a number of messages have been
669 queued up for a given host, without the overhead of a heavyweight queueing
673 <A NAME="13"><H2>15. Retries</H2></A>
676 When a message cannot immediately be directed, routed, or delivered, it
677 remains on the queue and another delivery attempt occurs at a later time.
678 While failures to deliver to remote hosts are the most common cause of this,
679 it is also possible for a message to be deferred as a result of temporary
680 local delivery failure, or following directing or routing. A local delivery
681 can fail if the user is over quota, while directing can be delayed if a user's
682 home directory is not available (e.g. missing NFS mount), and therefore the
683 existence of a .forward file cannot be tested. Routing can be delayed by DNS
687 Exim can be given a set of rules which specify how often to retry deferred
688 addresses, and when to give up. These rules apply to directing and routing as
689 well as to transporting, and are keyed by (wildcarded) domain name or, for
690 local users, by local-part and domain name, either of which can be wildcarded.
693 Each rule is actually a sequential list of subrules, which are applied
694 successively as time passes. At present there are two kinds of subrule: fixed
695 interval, and geometrically increasing interval. For example, it is possible
696 to specify a rule such as 'retry every 15 minutes for 2 hours; then increase
697 the interval between retries by a factor of 1.5 each time until 8 hours have
698 passed; then retry every 8 hours until 4 days have passed; then give up'. The
699 times are measured from when the address first failed, so, for example, if a
700 host has been down for 2 days, new messages will immediately go on to the
701 8-hour retry schedule.
704 Exim does not have an elaborate series of alarm clocks to cause retries to
705 happen exactly on schedule. A queue-runner process is started periodically, to
706 attempt delivery, one by one, of messages containing addresses that have
707 passed their next retry time. If such an address fails again, a new retry time
708 is computed, and so subsequent messages queued for the same address get
709 skipped. The queue is not processed sequentially, but in a 'random' order, to
710 prevent one rogue message that causes a problem blocking other messages to the
711 same destination for ever.
714 When the maximum time for retrying has passed, pending addresses are failed.
715 However, a next try time is still computed from the final subrule. Until that
716 time is reached, any new messages for the address are immediately failed. When
717 the next try time is passed, one further delivery attempt is made; if this
718 fails, a new next try time is computed, and so on.
721 The increasing number of small computers on the Internet has caused there to
722 be a lot of messages addressed to hosts that are never going to listen. The
723 retry logic described above should reduce the amount of wasted time spent on
724 trying to deliver such messages. However, some administrators are unhappy
725 about this rather draconian approach, which can cause an address to be failed
726 without any deliveries being attempted. Exim can alternatively be configured
727 always to try at least once those hosts whose last failure was before the
728 arrival of the message. This option increases the number of attempts to
729 deliver to dead hosts.
732 Retry rules can be predicated on particular errors as well as on domain names,
733 and for domains that are looked up in the DNS, further discrimination on
734 whether MX records were used or not is also possible. Thus it is possible to
735 treat 'connection refused' and 'connection timed out' differently, or to
736 distinguish between 'connection refused and there was only an A record' and
737 'connection refused from a host pointed to by an MX record'.
740 When a local delivery fails because a user is over quota, the retry rule can
741 be predicated on the length of time since the mailbox was last read. For
742 example, if the mailbox has been recently read, the delivery can be retried
743 for a while; otherwise it can be failed quickly.
746 <A NAME="14"><H2>16. Header rewriting</H2></A>
749 There are those who argue that header rewriting is a totally Bad Thing; there
750 are others who swear they cannot live without it. Exim provides the facility -
751 you do not have to use it!
754 Exim can be configured to rewrite the address portions of headers when a
755 message is received. For debugging purposes, the original headers are retained
756 in the spool file, but are not, of course, transported with the message.
757 Rewriting rules can be targeted at individual headers and the envelope fields;
758 it is possible, for example, just to rewrite the 'From' header and no others.
761 Rewriting rules are keyed by local-part and domain, either of which can be
762 wildcarded, and the replacement text is a general expansion string which can
763 contain file lookups. This makes it possible to replace login names by
764 'friendly' names in outgoing addresses via a DBM lookup, for example. The
765 other most common rewriting requirement of replacing *.foo.bar with foo.bar is
769 Headers are also automatically rewritten by Exim in two cases:
772 <LI> If a locally-generated message contains addresses without domains, a
773 configured qualifying domain is added to each of them. It is also
774 possible to specify which remote systems are permitted to send messages
775 containing unqualified addresses. These too get qualified on reception.
777 <LI> Routing of a domain may reveal that is was only a partial domain, in
778 which case the headers are rewritten to contain the full domain. For
779 example, as a result of routing, an address such as xxx@foo may turn into
783 <A NAME="15"><H2>17. Host verification</H2></A>
787 Exim can be configured to accept incoming SMTP calls from certain hosts only,
788 or it can be configured to reject calls from certain hosts. In both cases, the
789 test may include an RFC 1413 identification check. A system that gets all its
790 mail via a central hub might want to lock out the rest of the world, while a
791 number of systems under one management might want to exchange mail only via
792 the standard mailer, and hence reject mail from all but certain specified ids
796 When a host fails the acceptance test, Exim can either give an error code
797 immediately on connection, or allow the connection to proceed and then give
798 error codes to all the message's recipients. The latter approach is useful
799 when using the mechanism to reject unsolicited junk mail and mail bombs,
800 because it normally prevents the sender from trying again with the same
804 <A NAME="16"><H2>18. SMTP port reservation</H2></A>
807 The maximum number of simultaneous incoming SMTP calls can be set, and in
808 addition, a number of them can be reserved for particular hosts or particular
809 IP networks. It is also possible to specify a system load value above which
810 only calls from the reserved hosts are accepted.
813 <A NAME="17"><H2>19. Control of relaying</H2></A>
816 A host is said to act as a relay if it accepts an incoming message from an
817 external host and delivers it to an external host. Unscrupulous persons have
818 been known to use unsuspecting hosts as relays in an attempt to disguise the
819 origin of messages. An Exim host can be configured to accept mail from any
820 host for onward transmission to a specified set of domains only, and to accept
821 mail only from a specified list of hosts or networks for onward transmission
825 <A NAME="18"><H2>20. Sender verification</H2></A>
828 The return path of a message (also known as the 'envelope sender') is used
829 when Exim has to return an error message. If this is a bad address, the error
830 message cannot be delivered, and the postmaster has to sort things out.
833 Sender verification (a configurable option that applies to SMTP input) is
834 intended to pass this work to a foreign postmaster, by refusing to accept the
835 message in the first place. There is an exception list which can specify
836 certain hosts (with optional RFC 1413 identifications) that are allowed to
840 There are two main causes of bad return paths: misconfigured mailers (gateways
841 in particular), and users fooling around with mail. Sadly, the latter are
842 rather common in educational institutions. Sender verification catches both of
843 them. It operates by passing the sender address through the directors and
844 routers in verification mode; if this fails, the message is not accepted.
847 The first thing foreign postmasters ask when they learn about a rejected
848 message is 'What were the headers?'. For this reason, and also to collect
849 evidence in cases of mail forgery, Exim does not initially reject a message
850 after the MAIL FROM command in the SMTP session. It reads the message, so as
851 to be able to write the headers to the rejection log, and then gives a hard
852 error response to the sending host.
855 Unfortunately, several mailers believe that any error response after the data
856 for a message has been sent indicates a temporary error. Consequently, such
857 mailers will continue to try to send a message that has been rejected as
858 described above. To prevent this, whenever a message is rejected, Exim records
859 the time, bad address, and host in a DBM database. If the same host sends the
860 same bad address within 24 hours, it is rejected immediately at the MAIL FROM
864 Sadly, even this doesn't stop some mailers from repeatedly trying to send the
865 message. As a last resort, if the same host sends the same bad address for a
866 third time in 24 hours, the MAIL FROM command is accepted, but all subsequent
867 RCPT TO commands are rejected. If this does not stop a remote mailer then it
871 If the attempt to verify the sender address cannot be completed (typically
872 because of a DNS timeout) Exim gives temporary error code to the MAIL FROM
873 command, which should cause the remote mailer to try again later. However, it
874 is possible to configure Exim to accept the message in these circumstances.
877 Many messages with bad return paths in fact contain perfectly valid 'From' or
878 'Reply-to' headers. For administrators that want a quieter life, there is a
879 configuration option which causes Exim to check these headers if the return
880 path is bad, and if a good address is found, to use it to replace the return
881 path. The old value is retained in an X- header.
884 <A NAME="19"><H2>21. Sender lock out</H2></A>
887 More and more unsolicited junk mail is being seen on the Internet. It is
888 sometimes useful to be able to reject messages (from any host) with particular
889 sender addresses in the envelope. Exim can be configured to reject messages
890 whose sender addresses match certain patterns, either by failing the MAIL FROM
891 command, or (because some mailers take no notice of that) by failing all RCPT
895 <A NAME="20"><H2>22. Receiver verification</H2></A>
898 Exim can be configured so that it checks the addresses given in incoming SMTP
899 RCPT TO commands as they are received. A failing address can be immediately
900 rejected, or it can be logged and accepted. If verification cannot be
901 completed (typically because of a DNS timeout) either a temporary error code
902 can be given, or the address can be logged and accepted.
905 <A NAME="21"><H2>23. The 'percent hack'</H2></A>
908 The so-called 'percent hack' is the feature of mailers whereby a local-part
909 containing a percent sign gets interpreted as an entire new address, with the
910 percent replaced by @. This is used for explicit mail routing and sometimes
911 for testing. In Exim, it is possible to configure which local domains, if any,
912 allow the 'percent hack'.
915 <A NAME="22"><H2>24. Security</H2></A>
918 Exim is written as a single binary that has to run setuid to root. I did start
919 off trying to write it as a number of different modules, but soon came to the
920 conclusion that, for this type of mailer, it was not worth it, because the
921 functions don't decompose cleanly. For example, if you want to verify
922 addresses while receiving mail you need all the directing and routing
923 apparatus to be available.
926 Exim runs each local delivery in a separate process which is setuid to the
927 relevant local user. In addition, it can be configured to run under a given
928 non-root uid (and gid) for much of the rest of the time. In particular, it
929 need not be root while sending or receiving SMTP mail. On systems that do not
930 have the seteuid function, it uses setuid to give up root, which requires it
931 to re-invoke itself in order to regain the privilege when it needs to deliver
932 a message. On systems that do have seteuid, it can be configured to use that
933 function instead, thereby saving some resources.
936 Exim can be configured to use seteuid (on systems that have it) when reading a
937 .forward file in a user's home directory. This is necessary when home
938 directories are NFS mounted without root privilege, unless .forward files are
939 required to be world readable.
942 Exim checks the permissions and owners of files to which messages are to be
943 appended, and refuses to proceed with the delivery if things are not right.
946 Delivery of messages to pipes or files is supported only as a result of
947 expanding an address via an alias or a forward file, provided this is
948 permitted by the configuration. Externally generated local addresses cannot
949 specify files or pipes - no special action is taken for addresses starting
950 with the file or pipe characters, so they will usually fail.
953 Use of the VRFY function in SMTP connections is controlled by a configuration
954 option. The EXPN and DEBUG functions are not supported at all.
957 <A NAME="23"><H2>25. The Exim Monitor</H2></A>
960 A program for monitoring Exim and displaying information in an X window is
961 provided. This can be configured to show stripcharts of incoming and outgoing
962 mail in various categories. It also shows a 'tail' of the main log file, and
963 information about messages on the queue.
966 There is a menu of operations that can be performed by suitably privileged
967 users. Messages can be frozen, thawed, deleted, caused to be delivered,
968 modified, or returned to their senders from this interface.
git://git.exim.org / exim-website.git / blob