7bb17883c77143434880f96872197045d72b3a00
[exim.git] / doc / doc-txt / experimental-spec.txt
1 From time to time, experimental features may be added to Exim.
2 While a feature is experimental, there will be a build-time
3 option whose name starts "EXPERIMENTAL_" that must be set in
4 order to include the feature. This file contains information
5 about experimental features, all of which are unstable and
6 liable to incompatible change.
7
8
9 OCSP Stapling support
10 --------------------------------------------------------------
11
12 X.509 PKI certificates expire and can be revoked; to handle this, the
13 clients need some way to determine if a particular certificate, from a
14 particular Certificate Authority (CA), is still valid. There are three
15 main ways to do so.
16
17 The simplest way is to serve up a Certificate Revocation List (CRL) with
18 an ordinary web-server, regenerating the CRL before it expires. The
19 downside is that clients have to periodically re-download a potentially
20 huge file from every certificate authority it knows of.
21
22 The way with most moving parts at query time is Online Certificate
23 Status Protocol (OCSP), where the client verifies the certificate
24 against an OCSP server run by the CA. This lets the CA track all
25 usage of the certs. This requires running software with access to the
26 private key of the CA, to sign the responses to the OCSP queries. OCSP
27 is based on HTTP and can be proxied accordingly.
28
29 The only widespread OCSP server implementation (known to this writer)
30 comes as part of OpenSSL and aborts on an invalid request, such as
31 connecting to the port and then disconnecting. This requires
32 re-entering the passphrase each time some random client does this.
33
34 The third way is OCSP Stapling; in this, the server using a certificate
35 issued by the CA periodically requests an OCSP proof of validity from
36 the OCSP server, then serves it up inline as part of the TLS
37 negotiation. This approach adds no extra round trips, does not let the
38 CA track users, scales well with number of certs issued by the CA and is
39 resilient to temporary OCSP server failures, as long as the server
40 starts retrying to fetch an OCSP proof some time before its current
41 proof expires. The downside is that it requires server support.
42
43 If Exim is built with EXPERIMENTAL_OCSP and it was built with OpenSSL,
44 then it gains one new option: "tls_ocsp_file".
45
46 The file specified therein is expected to be in DER format, and contain
47 an OCSP proof. Exim will serve it as part of the TLS handshake. This
48 option will be re-expanded for SNI, if the tls_certificate option
49 contains $tls_sni, as per other TLS options.
50
51 Exim does not at this time implement any support for fetching a new OCSP
52 proof. The burden is on the administrator to handle this, outside of
53 Exim. The file specified should be replaced atomically, so that the
54 contents are always valid. Exim will expand the "tls_ocsp_file" option
55 on each connection, so a new file will be handled transparently on the
56 next connection.
57
58 Exim will check for a valid next update timestamp in the OCSP proof;
59 if not present, or if the proof has expired, it will be ignored.
60
61 At this point in time, we're gathering feedback on use, to determine if
62 it's worth adding complexity to the Exim daemon to periodically re-fetch
63 OCSP files and somehow handling multiple files. There is no client support
64 for OCSP in Exim, this is feature expected to be used by mail clients.
65
66
67
68
69 Brightmail AntiSpam (BMI) suppport
70 --------------------------------------------------------------
71
72 Brightmail AntiSpam is a commercial package. Please see
73 http://www.brightmail.com for more information on
74 the product. For the sake of clarity, we'll refer to it as
75 "BMI" from now on.
76
77
78 0) BMI concept and implementation overview
79
80 In contrast to how spam-scanning with SpamAssassin is
81 implemented in exiscan-acl, BMI is more suited for per
82 -recipient scanning of messages. However, each messages is
83 scanned only once, but multiple "verdicts" for multiple
84 recipients can be returned from the BMI server. The exiscan
85 implementation passes the message to the BMI server just
86 before accepting it. It then adds the retrieved verdicts to
87 the messages header file in the spool. These verdicts can then
88 be queried in routers, where operation is per-recipient
89 instead of per-message. To use BMI, you need to take the
90 following steps:
91
92 1) Compile Exim with BMI support
93 2) Set up main BMI options (top section of Exim config file)
94 3) Set up ACL control statement (ACL section of the config
95 file)
96 4) Set up your routers to use BMI verdicts (routers section
97 of the config file).
98 5) (Optional) Set up per-recipient opt-in information.
99
100 These four steps are explained in more details below.
101
102 1) Adding support for BMI at compile time
103
104 To compile with BMI support, you need to link Exim against
105 the Brighmail client SDK, consisting of a library
106 (libbmiclient_single.so) and a header file (bmi_api.h).
107 You'll also need to explicitly set a flag in the Makefile to
108 include BMI support in the Exim binary. Both can be achieved
109 with these lines in Local/Makefile:
110
111 EXPERIMENTAL_BRIGHTMAIL=yes
112 CFLAGS=-I/path/to/the/dir/with/the/includefile
113 EXTRALIBS_EXIM=-L/path/to/the/dir/with/the/library -lbmiclient_single
114
115 If you use other CFLAGS or EXTRALIBS_EXIM settings then
116 merge the content of these lines with them.
117
118 Note for BMI6.x users: You'll also have to add -lxml2_single
119 to the EXTRALIBS_EXIM line. Users of 5.5x do not need to do
120 this.
121
122 You should also include the location of
123 libbmiclient_single.so in your dynamic linker configuration
124 file (usually /etc/ld.so.conf) and run "ldconfig"
125 afterwards, or else the produced Exim binary will not be
126 able to find the library file.
127
128
129 2) Setting up BMI support in the Exim main configuration
130
131 To enable BMI support in the main Exim configuration, you
132 should set the path to the main BMI configuration file with
133 the "bmi_config_file" option, like this:
134
135 bmi_config_file = /opt/brightmail/etc/brightmail.cfg
136
137 This must go into section 1 of Exim's configuration file (You
138 can put it right on top). If you omit this option, it
139 defaults to /opt/brightmail/etc/brightmail.cfg.
140
141 Note for BMI6.x users: This file is in XML format in V6.xx
142 and its name is /opt/brightmail/etc/bmiconfig.xml. So BMI
143 6.x users MUST set the bmi_config_file option.
144
145
146 3) Set up ACL control statement
147
148 To optimize performance, it makes sense only to process
149 messages coming from remote, untrusted sources with the BMI
150 server. To set up a messages for processing by the BMI
151 server, you MUST set the "bmi_run" control statement in any
152 ACL for an incoming message. You will typically do this in
153 an "accept" block in the "acl_check_rcpt" ACL. You should
154 use the "accept" block(s) that accept messages from remote
155 servers for your own domain(s). Here is an example that uses
156 the "accept" blocks from Exim's default configuration file:
157
158
159 accept domains = +local_domains
160 endpass
161 verify = recipient
162 control = bmi_run
163
164 accept domains = +relay_to_domains
165 endpass
166 verify = recipient
167 control = bmi_run
168
169 If bmi_run is not set in any ACL during reception of the
170 message, it will NOT be passed to the BMI server.
171
172
173 4) Setting up routers to use BMI verdicts
174
175 When a message has been run through the BMI server, one or
176 more "verdicts" are present. Different recipients can have
177 different verdicts. Each recipient is treated individually
178 during routing, so you can query the verdicts by recipient
179 at that stage. From Exim's view, a verdict can have the
180 following outcomes:
181
182 o deliver the message normally
183 o deliver the message to an alternate location
184 o do not deliver the message
185
186 To query the verdict for a recipient, the implementation
187 offers the following tools:
188
189
190 - Boolean router preconditions. These can be used in any
191 router. For a simple implementation of BMI, these may be
192 all that you need. The following preconditions are
193 available:
194
195 o bmi_deliver_default
196
197 This precondition is TRUE if the verdict for the
198 recipient is to deliver the message normally. If the
199 message has not been processed by the BMI server, this
200 variable defaults to TRUE.
201
202 o bmi_deliver_alternate
203
204 This precondition is TRUE if the verdict for the
205 recipient is to deliver the message to an alternate
206 location. You can get the location string from the
207 $bmi_alt_location expansion variable if you need it. See
208 further below. If the message has not been processed by
209 the BMI server, this variable defaults to FALSE.
210
211 o bmi_dont_deliver
212
213 This precondition is TRUE if the verdict for the
214 recipient is NOT to deliver the message to the
215 recipient. You will typically use this precondition in a
216 top-level blackhole router, like this:
217
218 # don't deliver messages handled by the BMI server
219 bmi_blackhole:
220 driver = redirect
221 bmi_dont_deliver
222 data = :blackhole:
223
224 This router should be on top of all others, so messages
225 that should not be delivered do not reach other routers
226 at all. If the message has not been processed by
227 the BMI server, this variable defaults to FALSE.
228
229
230 - A list router precondition to query if rules "fired" on
231 the message for the recipient. Its name is "bmi_rule". You
232 use it by passing it a colon-separated list of rule
233 numbers. You can use this condition to route messages that
234 matched specific rules. Here is an example:
235
236 # special router for BMI rule #5, #8 and #11
237 bmi_rule_redirect:
238 driver = redirect
239 bmi_rule = 5:8:11
240 data = postmaster@mydomain.com
241
242
243 - Expansion variables. Several expansion variables are set
244 during routing. You can use them in custom router
245 conditions, for example. The following variables are
246 available:
247
248 o $bmi_base64_verdict
249
250 This variable will contain the BASE64 encoded verdict
251 for the recipient being routed. You can use it to add a
252 header to messages for tracking purposes, for example:
253
254 localuser:
255 driver = accept
256 check_local_user
257 headers_add = X-Brightmail-Verdict: $bmi_base64_verdict
258 transport = local_delivery
259
260 If there is no verdict available for the recipient being
261 routed, this variable contains the empty string.
262
263 o $bmi_base64_tracker_verdict
264
265 This variable will contain a BASE64 encoded subset of
266 the verdict information concerning the "rules" that
267 fired on the message. You can add this string to a
268 header, commonly named "X-Brightmail-Tracker". Example:
269
270 localuser:
271 driver = accept
272 check_local_user
273 headers_add = X-Brightmail-Tracker: $bmi_base64_tracker_verdict
274 transport = local_delivery
275
276 If there is no verdict available for the recipient being
277 routed, this variable contains the empty string.
278
279 o $bmi_alt_location
280
281 If the verdict is to redirect the message to an
282 alternate location, this variable will contain the
283 alternate location string returned by the BMI server. In
284 its default configuration, this is a header-like string
285 that can be added to the message with "headers_add". If
286 there is no verdict available for the recipient being
287 routed, or if the message is to be delivered normally,
288 this variable contains the empty string.
289
290 o $bmi_deliver
291
292 This is an additional integer variable that can be used
293 to query if the message should be delivered at all. You
294 should use router preconditions instead if possible.
295
296 $bmi_deliver is '0': the message should NOT be delivered.
297 $bmi_deliver is '1': the message should be delivered.
298
299
300 IMPORTANT NOTE: Verdict inheritance.
301 The message is passed to the BMI server during message
302 reception, using the target addresses from the RCPT TO:
303 commands in the SMTP transaction. If recipients get expanded
304 or re-written (for example by aliasing), the new address(es)
305 inherit the verdict from the original address. This means
306 that verdicts also apply to all "child" addresses generated
307 from top-level addresses that were sent to the BMI server.
308
309
310 5) Using per-recipient opt-in information (Optional)
311
312 The BMI server features multiple scanning "profiles" for
313 individual recipients. These are usually stored in a LDAP
314 server and are queried by the BMI server itself. However,
315 you can also pass opt-in data for each recipient from the
316 MTA to the BMI server. This is particularly useful if you
317 already look up recipient data in Exim anyway (which can
318 also be stored in a SQL database or other source). This
319 implementation enables you to pass opt-in data to the BMI
320 server in the RCPT ACL. This works by setting the
321 'bmi_optin' modifier in a block of that ACL. If should be
322 set to a list of comma-separated strings that identify the
323 features which the BMI server should use for that particular
324 recipient. Ideally, you would use the 'bmi_optin' modifier
325 in the same ACL block where you set the 'bmi_run' control
326 flag. Here is an example that will pull opt-in data for each
327 recipient from a flat file called
328 '/etc/exim/bmi_optin_data'.
329
330 The file format:
331
332 user1@mydomain.com: <OPTIN STRING1>:<OPTIN STRING2>
333 user2@thatdomain.com: <OPTIN STRING3>
334
335
336 The example:
337
338 accept domains = +relay_to_domains
339 endpass
340 verify = recipient
341 bmi_optin = ${lookup{$local_part@$domain}lsearch{/etc/exim/bmi_optin_data}}
342 control = bmi_run
343
344 Of course, you can also use any other lookup method that
345 Exim supports, including LDAP, Postgres, MySQL, Oracle etc.,
346 as long as the result is a list of colon-separated opt-in
347 strings.
348
349 For a list of available opt-in strings, please contact your
350 Brightmail representative.
351
352
353
354
355 Sender Policy Framework (SPF) support
356 --------------------------------------------------------------
357
358 To learn more about SPF, visit http://www.openspf.org. This
359 document does not explain the SPF fundamentals, you should
360 read and understand the implications of deploying SPF on your
361 system before doing so.
362
363 SPF support is added via the libspf2 library. Visit
364
365 http://www.libspf2.org/
366
367 to obtain a copy, then compile and install it. By default,
368 this will put headers in /usr/local/include and the static
369 library in /usr/local/lib.
370
371 To compile Exim with SPF support, set these additional flags in
372 Local/Makefile:
373
374 EXPERIMENTAL_SPF=yes
375 CFLAGS=-DSPF -I/usr/local/include
376 EXTRALIBS_EXIM=-L/usr/local/lib -lspf2
377
378 This assumes that the libspf2 files are installed in
379 their default locations.
380
381 You can now run SPF checks in incoming SMTP by using the "spf"
382 ACL condition in either the MAIL, RCPT or DATA ACLs. When
383 using it in the RCPT ACL, you can make the checks dependent on
384 the RCPT address (or domain), so you can check SPF records
385 only for certain target domains. This gives you the
386 possibility to opt-out certain customers that do not want
387 their mail to be subject to SPF checking.
388
389 The spf condition takes a list of strings on its right-hand
390 side. These strings describe the outcome of the SPF check for
391 which the spf condition should succeed. Valid strings are:
392
393 o pass The SPF check passed, the sending host
394 is positively verified by SPF.
395 o fail The SPF check failed, the sending host
396 is NOT allowed to send mail for the domain
397 in the envelope-from address.
398 o softfail The SPF check failed, but the queried
399 domain can't absolutely confirm that this
400 is a forgery.
401 o none The queried domain does not publish SPF
402 records.
403 o neutral The SPF check returned a "neutral" state.
404 This means the queried domain has published
405 a SPF record, but wants to allow outside
406 servers to send mail under its domain as well.
407 o err_perm This indicates a syntax error in the SPF
408 record of the queried domain. This should be
409 treated like "none".
410 o err_temp This indicates a temporary error during all
411 processing, including Exim's SPF processing.
412 You may defer messages when this occurs.
413
414 You can prefix each string with an exclamation mark to invert
415 is meaning, for example "!fail" will match all results but
416 "fail". The string list is evaluated left-to-right, in a
417 short-circuit fashion. When a string matches the outcome of
418 the SPF check, the condition succeeds. If none of the listed
419 strings matches the outcome of the SPF check, the condition
420 fails.
421
422 Here is an example to fail forgery attempts from domains that
423 publish SPF records:
424
425 /* -----------------
426 deny message = $sender_host_address is not allowed to send mail from ${if def:sender_address_domain {$sender_address_domain}{$sender_helo_name}}. \
427 Please see http://www.openspf.org/Why?scope=${if def:sender_address_domain {mfrom}{helo}};identity=${if def:sender_address_domain {$sender_address}{$sender_helo_name}};ip=$sender_host_address
428 spf = fail
429 --------------------- */
430
431 You can also give special treatment to specific domains:
432
433 /* -----------------
434 deny message = AOL sender, but not from AOL-approved relay.
435 sender_domains = aol.com
436 spf = fail:neutral
437 --------------------- */
438
439 Explanation: AOL publishes SPF records, but is liberal and
440 still allows non-approved relays to send mail from aol.com.
441 This will result in a "neutral" state, while mail from genuine
442 AOL servers will result in "pass". The example above takes
443 this into account and treats "neutral" like "fail", but only
444 for aol.com. Please note that this violates the SPF draft.
445
446 When the spf condition has run, it sets up several expansion
447 variables.
448
449 $spf_header_comment
450 This contains a human-readable string describing the outcome
451 of the SPF check. You can add it to a custom header or use
452 it for logging purposes.
453
454 $spf_received
455 This contains a complete Received-SPF: header that can be
456 added to the message. Please note that according to the SPF
457 draft, this header must be added at the top of the header
458 list. Please see section 10 on how you can do this.
459
460 Note: in case of "Best-guess" (see below), the convention is
461 to put this string in a header called X-SPF-Guess: instead.
462
463 $spf_result
464 This contains the outcome of the SPF check in string form,
465 one of pass, fail, softfail, none, neutral, err_perm or
466 err_temp.
467
468 $spf_smtp_comment
469 This contains a string that can be used in a SMTP response
470 to the calling party. Useful for "fail".
471
472 In addition to SPF, you can also perform checks for so-called
473 "Best-guess". Strictly speaking, "Best-guess" is not standard
474 SPF, but it is supported by the same framework that enables SPF
475 capability. Refer to http://www.openspf.org/FAQ/Best_guess_record
476 for a description of what it means.
477
478 To access this feature, simply use the spf_guess condition in place
479 of the spf one. For example:
480
481 /* -----------------
482 deny message = $sender_host_address doesn't look trustworthy to me
483 spf_guess = fail
484 --------------------- */
485
486 In case you decide to reject messages based on this check, you
487 should note that although it uses the same framework, "Best-guess"
488 is NOT SPF, and therefore you should not mention SPF at all in your
489 reject message.
490
491 When the spf_guess condition has run, it sets up the same expansion
492 variables as when spf condition is run, described above.
493
494 Additionally, since Best-guess is not standardized, you may redefine
495 what "Best-guess" means to you by redefining spf_guess variable in
496 global config. For example, the following:
497
498 /* -----------------
499 spf_guess = v=spf1 a/16 mx/16 ptr ?all
500 --------------------- */
501
502 would relax host matching rules to a broader network range.
503
504
505 SRS (Sender Rewriting Scheme) Support
506 --------------------------------------------------------------
507
508 Exiscan currently includes SRS support via Miles Wilton's
509 libsrs_alt library. The current version of the supported
510 library is 0.5.
511
512 In order to use SRS, you must get a copy of libsrs_alt from
513
514 http://srs.mirtol.com/
515
516 Unpack the tarball, then refer to MTAs/README.EXIM
517 to proceed. You need to set
518
519 EXPERIMENTAL_SRS=yes
520
521 in your Local/Makefile.
522
523
524 DCC Support
525 --------------------------------------------------------------
526
527 *) Building exim
528
529 In order to build exim with DCC support add
530
531 EXPERIMENTAL_DCC=yes
532
533 to your Makefile. (Re-)build/install exim. exim -d should show
534 EXPERIMENTAL_DCC under "Support for".
535
536
537 *) Configuration
538
539 In the main section of exim.cf add at least
540 dccifd_address = /usr/local/dcc/var/dccifd
541 or
542 dccifd_address = <ip> <port>
543
544 In the DATA ACL you can use the new condition
545 dcc = *
546
547 After that "$dcc_header" contains the X-DCC-Header.
548
549 Return values are:
550 fail for overall "R", "G" from dccifd
551 defer for overall "T" from dccifd
552 accept for overall "A", "S" from dccifd
553
554 dcc = */defer_ok works as for spamd.
555
556 The "$dcc_result" variable contains the overall result from DCC
557 answer. There will an X-DCC: header added to the mail.
558
559 Usually you'll use
560 defer !dcc = *
561 to greylist with DCC.
562
563 If you set, in the main section,
564 dcc_direct_add_header = true
565 then the dcc header will be added "in deep" and if the spool
566 file was already written it gets removed. This forces Exim to
567 write it again if needed. This helps to get the DCC Header
568 through to eg. SpamAssassin.
569
570 If you want to pass even more headers in the middle of the
571 DATA stage you can set
572 $acl_m_dcc_add_header
573 to tell the DCC routines add more information; eg, you might set
574 this to some results from ClamAV. Be careful. Header syntax is
575 not checked and is added "as is".
576
577
578 --------------------------------------------------------------
579 End of file
580 --------------------------------------------------------------