1 .TH SLAPD-META 5 "RELEASEDATE" "OpenLDAP LDVERSION"
2 .\" Copyright 1998-2003 The OpenLDAP Foundation, All Rights Reserved.
3 .\" Copying restrictions apply. See the COPYRIGHT file.
4 .\" Copyright 2001, Pierangelo Masarati, All rights reserved. <ando@sys-net.it>
7 .\" Portions of this document should probably be moved to slapd-ldap(5)
8 .\" and maybe manual pages for librewrite.
11 slapd-meta \- metadirectory backend
19 performs basic LDAP proxying with respect to a set of remote LDAP
20 servers, called "targets".
21 The information contained in these servers can be presented as
22 belonging to a single Directory Information Tree (DIT).
24 A basic knowledge of the functionality of the
26 backend is recommended.
27 This backend has been designed as an enhancement of the ldap backend.
28 The two backends share many features (actually they also share
32 backend is intended to proxy operations directed to a single server, the
34 backend is mainly intended for proxying of multiple servers and possibly
35 naming context masquerading.
36 These features, although useful in many scenarios, may result in
37 excessive overhead for some applications, so its use should be
39 In the examples section, some typical scenarios will be discussed.
41 There are examples in various places in this document, as well as in the
42 slapd/back-meta/data/ directory in the OpenLDAP source tree.
46 options apply to the META backend database.
47 That is, they must follow a "database meta" line and come before any
48 subsequent "backend" or "database" lines.
49 Other database options are described in the
55 backend, operational attributes related to entry creation/modification
56 should not be used, as they would be passed to the target servers,
58 Moreover, it makes little sense to use such attributes in proxying, as
59 the proxy server doesn't actually store data, so it should have no
60 knowledge of such attributes.
61 While code to strip the modification attributes has been put in place
62 (and #ifdef'd), it implies unmotivated overhead.
63 So it is strongly recommended to set
72 .SH SPECIAL CONFIGURATION DIRECTIVES
73 Target configuration starts with the "uri" directive.
74 All the configuration directives that are not specific to targets
75 should be defined first for clarity, including those that are common
79 .B default-target none
80 This directive forces the backend to reject all those operations
81 that must resolve to a single target in case none or multiple
83 They include: add, delete, modify, modrdn; compare is not included, as
84 well as bind since, as they don't alter entries, in case of multiple
85 matches an attempt is made to perform the operation on any candidate
86 target, with the constraint that at most one must succeed.
87 This directive can also be used when processing targets to mark a
88 specific target as default.
90 .B dncache-ttl {forever|disabled|<ttl>}
91 This directive sets the time-to-live of the DN cache.
92 This caches the target that holds a given DN to speed up target
93 selection in case multiple targets would result from an uncached
94 search; forever means cache never expires; disabled means no DN
95 caching; otherwise a valid ( > 0 ) ttl in seconds is required.
96 .SH TARGET SPECIFICATION
97 Target specification starts with a "uri" directive:
99 .B uri <protocol>://[<host>[:<port>]]/<naming context>
100 The "server" directive that was allowed in the LDAP backend (although
101 deprecated) has been discarded in the Meta backend.
102 The <protocol> part can be anything
103 .BR ldap_initialize (3)
104 accepts ({ldap|ldaps|ldapi} and variants); <host> and <port> may be
105 omitted, defaulting to whatever is set in /etc/ldap.conf.
106 The <naming context> part is mandatory.
107 It must end with one of the naming contexts defined for the backend,
112 suffix "\fBdc=foo,dc=com\fP"
113 uri "ldap://x.foo.com/dc=x,\fBdc=foo,dc=com\fP"
117 The <naming context> part doesn't need to be unique across the targets;
118 it may also match one of the values of the "suffix" directive.
119 Multiple URIs may be defined in a single argument. The URIs must
120 be separated by TABs (e.g. '\\t'), and the additional URIs must have
121 no <naming context> part. This causes the underlying library
122 to contact the first server of the list that responds.
124 .B default-target [<target>]
125 The "default-target" directive can also be used during target specification.
126 With no arguments it marks the current target as the default.
127 The optional number marks target <target> as the default one, starting
129 Target <target> must be defined.
131 .B binddn "<administrative DN for access control purposes>"
132 This directive, as in the LDAP backend, allows to define the DN that is
133 used to query the target server for acl checking; it should have read
134 access on the target server to attributes used on the proxy for acl
136 There is no risk of giving away such values; they are only used to
139 .B bindpw <password for access control purposes>
140 This directive sets the password for acl checking in conjunction
141 with the above mentioned "binddn" directive.
144 If this option is given, the client's bind credentials are remembered
145 for rebinds when chasing referrals.
147 .B pseudorootdn "<substitute DN in case of rootdn bind>"
148 This directive, if present, sets the DN that will be substituted to
149 the bind DN if a bind with the backend's "rootdn" succeeds.
150 The true "rootdn" of the target server ought not be used; an arbitrary
151 administrative DN should used instead.
153 .B pseudorootpw "<substitute password in case of rootdn bind>"
154 This directive sets the credential that will be used in case a bind
155 with the backend's "rootdn" succeeds, and the bind is propagated to
156 the target using the "pseudorootdn" DN.
158 Note: cleartext credentials must be supplied here; as a consequence,
159 using the pseudorootdn/pseudorootpw directives is inherently unsafe.
162 The rewrite options are described in the "REWRITING" section.
164 .B suffixmassage "<virtual naming context>" "<real naming context>"
165 All the directives starting with "rewrite" refer to the rewrite engine
166 that has been added to slapd.
167 The "suffixmassage" directive was introduced in the LDAP backend to
168 allow suffix massaging while proxying.
169 It has been obsoleted by the rewriting tools.
170 However, both for backward compatibility and for ease of configuration
171 when simple suffix massage is required, it has been preserved.
172 It wraps the basic rewriting instructions that perform suffix
175 Note: this also fixes a flaw in suffix massaging, which operated
176 on (case insensitive) DNs instead of normalized DNs,
177 so "dc=foo, dc=com" would not match "dc=foo,dc=com".
179 See the "REWRITING" section.
181 .B map "{attribute|objectclass} [<local name>|*] {<foreign name>|*}"
182 This maps object classes and attributes as in the LDAP backend.
186 A powerful (and in some sense dangerous) rewrite engine has been added
187 to both the LDAP and Meta backends.
188 While the former can gain limited beneficial effects from rewriting
189 stuff, the latter can become an amazingly powerful tool.
191 Consider a couple of scenarios first.
193 1) Two directory servers share two levels of naming context;
194 say "dc=a,dc=foo,dc=com" and "dc=b,dc=foo,dc=com".
195 Then, an unambiguous Meta database can be configured as:
200 suffix "\fBdc=foo,dc=com\fP"
201 uri "ldap://a.foo.com/dc=a,\fBdc=foo,dc=com\fP"
202 uri "ldap://b.foo.com/dc=b,\fBdc=foo,dc=com\fP"
206 Operations directed to a specific target can be easily resolved
207 because there are no ambiguities.
208 The only operation that may resolve to multiple targets is a search
209 with base "dc=foo,dc=com" and scope at least "one", which results in
210 spawning two searches to the targets.
212 2a) Two directory servers don't share any portion of naming context,
213 but they'd present as a single DIT
214 [Caveat: uniqueness of (massaged) entries among the two servers is
215 assumed; integrity checks risk to incur in excessive overhead and have
216 not been implemented].
217 Say we have "dc=bar,dc=org" and "o=Foo,c=US",
218 and we'd like them to appear as branches of "dc=foo,dc=com", say
219 "dc=a,dc=foo,dc=com" and "dc=b,dc=foo,dc=com".
220 Then we need to configure our Meta backend as:
225 suffix "dc=foo,dc=com"
227 uri "ldap://a.bar.com/\fBdc=a,dc=foo,dc=com\fP"
228 suffixmassage "\fBdc=a,dc=foo,dc=com\fP" "dc=bar,dc=org"
230 uri "ldap://b.foo.com/\fBdc=b,dc=foo,dc=com\fP"
231 suffixmassage "\fBdc=b,dc=foo,dc=com\fP" "o=Foo,c=US"
235 Again, operations can be resolved without ambiguity, although
236 some rewriting is required.
237 Notice that the virtual naming context of each target is a branch of
238 the database's naming context; it is rewritten back and forth when
239 operations are performed towards the target servers.
240 What "back and forth" means will be clarified later.
242 When a search with base "dc=foo,dc=com" is attempted, if the
243 scope is "base" it fails with "no such object"; in fact, the
244 common root of the two targets (prior to massaging) does not
246 If the scope is "one", both targets are contacted with the base
247 replaced by each target's base; the scope is derated to "base".
248 In general, a scope "one" search is honored, and the scope is derated,
249 only when the incoming base is at most one level lower of a target's
250 naming context (prior to massaging).
252 Finally, if the scope is "sub" the incoming base is replaced
253 by each target's unmassaged naming context, and the scope
256 2b) Consider the above reported scenario with the two servers
257 sharing the same naming context:
262 suffix "\fBdc=foo,dc=com\fP"
264 uri "ldap://a.bar.com/\fBdc=foo,dc=com\fP"
265 suffixmassage "\fBdc=foo,dc=com\fP" "dc=bar,dc=org"
267 uri "ldap://b.foo.com/\fBdc=foo,dc=com\fP"
268 suffixmassage "\fBdc=foo,dc=com\fP" "o=Foo,c=US"
272 All the previous considerations hold, except that now there is
273 no way to unambiguously resolve a DN.
274 In this case, all the operations that require an unambiguous target
275 selection will fail unless the DN is already cached or a default
277 Practical configurations may result as a combination of all the
280 Note on ACLs: at present you may add whatever ACL rule you desire
281 to to the Meta (and LDAP) backends.
282 However, the meaning of an ACL on a proxy may require some
284 Two philosophies may be considered:
286 a) the remote server dictates the permissions; the proxy simply passes
287 back what it gets from the remote server.
289 b) the remote server unveils "everything"; the proxy is responsible
290 for protecting data from unauthorized access.
292 Of course the latter sounds unreasonable, but it is not.
293 It is possible to imagine scenarios in which a remote host discloses
294 data that can be considered "public" inside an intranet, and a proxy
295 that connects it to the internet may impose additional constraints.
296 To this purpose, the proxy should be able to comply with all the ACL
297 matching criteria that the server supports.
298 This has been achieved with regard to all the criteria supported by
299 slapd except a special subtle case (please drop me a note if you can
300 find other exceptions: <ando@openldap.org>).
305 access to dn="<dn>" attr=<attr>
306 by dnattr=<dnattr> read
311 cannot be matched iff the attribute that is being requested, <attr>,
312 is NOT <dnattr>, and the attribute that determines membership,
313 <dnattr>, has not been requested (e.g. in a search)
315 In fact this ACL is resolved by slapd using the portion of entry it
316 retrieved from the remote server without requiring any further
317 intervention of the backend, so, if the <dnattr> attribute has not
318 been fetched, the match cannot be assessed because the attribute is
319 not present, not because no value matches the requirement!
321 Note on ACLs and attribute mapping: ACLs are applied to the mapped
322 attributes; for instance, if the attribute locally known as "foo" is
323 mapped to "bar" on a remote server, then local ACLs apply to attribute
324 "foo" and are totally unaware of its remote name.
325 The remote server will check permissions for "bar", and the local
326 server will possibly enforce additional restrictions to "foo".
328 .\" If this section is moved, also update the reference in
329 .\" libraries/librewrite/RATIONALE.
332 A string is rewritten according to a set of rules, called a `rewrite
334 The rules are based on Regular Expressions (POSIX regex) with
335 substring matching; extensions are planned to allow basic variable
336 substitution and map resolution of substrings.
337 The behavior of pattern matching/substitution can be altered by a set
340 The underlying concept is to build a lightweight rewrite module
341 for the slapd server (initially dedicated to the LDAP backend).
343 An incoming string is matched agains a set of rules.
344 Rules are made of a match pattern, a substitution pattern and a set of
346 In case of match a string rewriting is performed according to the
347 substitution pattern that allows to refer to substrings matched in the
349 The actions, if any, are finally performed.
350 The substitution pattern allows map resolution of substrings.
351 A map is a generic object that maps a substitution pattern to a value.
352 .SH "Pattern Matching Flags"
355 honors case in matching (default is case insensitive)
358 use POSIX Basic Regular Expressions (default is Extended)
362 apply the rule once only (default is recursive)
365 stop applying rules in case of match.
368 stop current operation if the rule matches, and issue an `unwilling to
372 jump n rules back and forth (watch for loops!).
373 Note that `G{1}' is implicit in every rule.
376 ignores errors in rule; this means, in case of error, e.g. issued by a
377 map, the error is treated as a missed match.
378 The `unwilling to perform' is not overridden.
380 The ordering of the flags is significant.
381 For instance: `IG{2}' means ignore errors and jump two lines ahead
382 both in case of match and in case of error, while `G{2}I' means ignore
383 errors, but jump thwo lines ahead only in case of match.
385 More flags (mainly Action Flags) will be added as needed.
386 .SH "Pattern matching:"
389 .SH "Substitution Pattern Syntax:"
390 Everything starting with `%' requires substitution;
392 the only obvious exception is `%%', which is left as is;
394 the basic substitution is `%d', where `d' is a digit;
395 0 means the whole string, while 1-9 is a submatch, as discussed in
398 a `%' followed by a `{' invokes an advanced substitution.
402 `%' `{' [ <op> ] <name> `(' <substitution> `)' `}'
405 where <name> must be a legal name for the map, i.e.
409 <name> ::= [a-z][a-z0-9]* (case insensitive)
410 <op> ::= `>' `|' `&' `&&' `*' `**' `$'
414 and <substitution> must be a legal substitution
415 pattern, with no limits on the nesting level.
420 sub context invocation; <name> must be a legal, already defined
424 external command invocation; <name> must refer to a legal, already
425 defined command name (NOT IMPL.)
428 variable assignment; <name> defines a variable in the running
429 operation structure which can be dereferenced later; operator
431 assigns a variable in the rewrite context scope; operator
433 assigns a variable that scopes the entire session, e.g. its value
434 can be derefenced later by other rewrite contexts
437 variable dereferencing; <name> must refer to a variable that is
438 defined and assigned for the running operation; operator
440 dereferences a variable scoping the rewrite context; operator
442 dereferences a variable scoping the whole session, e.g. the value
443 is passed across rewrite contexts
446 parameter dereferencing; <name> must refer to an existing parameter;
447 the idea is to make some run-time parameters set by the system
448 available to the rewrite engine, as the client host name, the bind DN
449 if any, constant parameters initialized at config time, and so on;
450 no parameter is currently set by either
454 but constant parameters can be defined in the configuration file
459 Substitution escaping has been delegated to the `%' symbol,
460 which is used instead of `\e' in string substitution patterns
461 because `\e' is already escaped by slapd's low level parsing routines;
464 escaping requires two `\e' symbols, e.g. `\fB.*\e.foo\e.bar\fP' must
465 be written as `\fB.*\e\e.foo\e\e.bar\fP'.
467 .\" The symbol can be altered at will by redefining the related macro in
470 .SH "Rewrite context:"
471 A rewrite context is a set of rules which are applied in sequence.
472 The basic idea is to have an application initialize a rewrite
473 engine (think of Apache's mod_rewrite ...) with a set of rewrite
474 contexts; when string rewriting is required, one invokes the
475 appropriate rewrite context with the input string and obtains the
476 newly rewritten one if no errors occur.
478 Each basic server operation is associated to a rewrite context;
479 they are divided in two main groups: client \-> server and
480 server \-> client rewriting.
486 (default) if defined and no specific context
492 compareAttrDN compare AVA
496 modifyAttrDN modify AVA
507 searchResult search (only if defined; no default;
508 acts on DN and DN-syntax attributes
510 searchAttrDN search AVA
511 matchedDN all ops (only if applicable)
515 .SH "Basic configuration syntax"
517 .B rewriteEngine { on | off }
518 If `on', the requested rewriting is performed; if `off', no
519 rewriting takes place (an easy way to stop rewriting without
520 altering too much the configuration file).
522 .B rewriteContext <context name> "[ alias <aliased context name> ]"
523 <Context name> is the name that identifies the context, i.e. the name
524 used by the application to refer to the set of rules it contains.
525 It is used also to reference sub contexts in string rewriting.
526 A context may aliase another one.
527 In this case the alias context contains no rule, and any reference to
528 it will result in accessing the aliased one.
530 .B rewriteRule "<regex pattern>" "<substitution pattern>" "[ <flags> ]"
531 Determines how a tring can be rewritten if a pattern is matched.
532 Examples are reported below.
533 .SH "Additional configuration syntax:"
535 .B rewriteMap "<map name>" "<map type>" "[ <map attrs> ]"
536 Allows to define a map that transforms substring rewriting into
538 The map is referenced inside the substitution pattern of a rule.
540 .B rewriteParam <param name> <param value>
541 Sets a value with global scope, that can be dereferenced by the
542 command `%{$paramName}'.
544 .B rewriteMaxPasses <number of passes>
545 Sets the maximum number of total rewriting passes that can be
546 performed in a single rewrite operation (to avoid loops).
547 .SH "Configuration examples:"
549 # set to `off' to disable rewriting
552 # Everything defined here goes into the `default' context.
553 # This rule changes the naming context of anything sent
554 # to `dc=home,dc=net' to `dc=OpenLDAP, dc=org'
556 rewriteRule "(.*)dc=home,[ ]?dc=net"
557 "%1dc=OpenLDAP, dc=org" ":"
559 # since a pretty/normalized DN does not include spaces
560 # after rdn separators, e.g. `,', this rule suffices:
562 rewriteRule "(.*)dc=home,dc=net"
563 "%1dc=OpenLDAP,dc=org" ":"
565 # Start a new context (ends input of the previous one).
566 # This rule adds blanks between DN parts if not present.
567 rewriteContext addBlanks
568 rewriteRule "(.*),([^ ].*)" "%1, %2"
570 # This one eats blanks
571 rewriteContext eatBlanks
572 rewriteRule "(.*),[ ](.*)" "%1,%2"
574 # Here control goes back to the default rewrite
575 # context; rules are appended to the existing ones.
576 # anything that gets here is piped into rule `addBlanks'
577 rewriteContext default
578 rewriteRule ".*" "%{>addBlanks(%0)}" ":"
580 .\" # Anything with `uid=username' is looked up in
581 .\" # /etc/passwd for gecos (I know it's nearly useless,
582 .\" # but it is there just as a guideline to implementing
584 .\" # Note the `I' flag that leaves `uid=username' in place
585 .\" # if `username' does not have a valid account, and the
586 .\" # `:' that forces the rule to be processed exactly once.
587 .\" rewriteContext uid2Gecos
588 .\" rewriteRule "(.*)uid=([a-z0-9]+),(.+)"
589 .\" "%1cn=%2{xpasswd},%3" "I:"
591 .\" # Finally, in a bind, if one uses a `uid=username' DN,
592 .\" # it is rewritten in `cn=name surname' if possible.
593 .\" rewriteContext bindDn
594 .\" rewriteRule ".*" "%{>addBlanks(%{>uid2Gecos(%0)})}" ":"
596 # Rewrite the search base according to `default' rules.
597 rewriteContext searchBase alias default
599 # Search results with OpenLDAP DN are rewritten back with
600 # `dc=home,dc=net' naming context, with spaces eaten.
601 rewriteContext searchResult
602 rewriteRule "(.*[^ ]?)[ ]?dc=OpenLDAP,[ ]?dc=org"
603 "%{>eatBlanks(%1)}dc=home,dc=net" ":"
605 # Bind with email instead of full DN: we first need
606 # an ldap map that turns attributes into a DN (the
607 # argument used when invoking the map is appended to
608 # the URI and acts as the filter portion)
609 rewriteMap ldap attr2dn "ldap://host/dc=my,dc=org?dn?sub"
611 # Then we need to detect DN made up of a single email,
612 # e.g. `mail=someone@example.com'; note that the rule
613 # in case of match stops rewriting; in case of error,
614 # it is ignored. In case we are mapping virtual
615 # to real naming contexts, we also need to rewrite
616 # regular DNs, because the definition of a bindDn
617 # rewrite context overrides the default definition.
618 rewriteContext bindDn
619 rewriteRule "^mail=[^,]+@[^,]+$" "%{attr2dn(%0)}" "@I"
621 # This is a rather sophisticated example. It massages a
622 # search filter in case who performs the search has
623 # administrative privileges. First we need to keep
624 # track of the bind DN of the incoming request, which is
625 # stored in a variable called `binddn' with session scope,
626 # and left in place to allow regular binding:
627 rewriteContext bindDn
628 rewriteRule ".+" "%{&&binddn(%0)}%0" ":"
630 # A search filter containing `uid=' is rewritten only
631 # if an appropriate DN is bound.
632 # To do this, in the first rule the bound DN is
633 # dereferenced, while the filter is decomposed in a
634 # prefix, in the value of the `uid=<arg>' AVA, and
635 # in a suffix. A tag `<>' is appended to the DN.
636 # If the DN refers to an entry in the `ou=admin' subtree,
637 # the filter is rewritten OR-ing the `uid=<arg>' with
638 # `cn=<arg>'; otherwise it is left as is. This could be
639 # useful, for instance, to allow apache's auth_ldap-1.4
640 # module to authenticate users with both `uid' and
641 # `cn', but only if the request comes from a possible
642 # `cn=Web auth,ou=admin,dc=home,dc=net' user.
643 rewriteContext searchFilter
644 rewriteRule "(.*\e\e()uid=([a-z0-9_]+)(\e\e).*)"
645 "%{**binddn}<>%{&prefix(%1)}%{&arg(%2)}%{&suffix(%3)}"
647 rewriteRule "[^,]+,ou=admin,dc=home,dc=net"
648 "%{*prefix}|(uid=%{*arg})(cn=%{*arg})%{*suffix}" "@I"
649 rewriteRule ".*<>" "%{*prefix}uid=%{*arg}%{*suffix}" ":"
651 # This example shows how to strip unwanted DN-valued
652 # attribute values from a search result; the first rule
653 # matches DN values below "ou=People,dc=example,dc=com";
654 # in case of match the rewriting exits successfully.
655 # The second rule matches everything else and causes
656 # the value to be rejected.
657 rewriteContext searchResult
658 rewriteRule ".*,ou=People,dc=example,dc=com" "%0" "@"
659 rewriteRule ".*" "" "#"
661 .SH "LDAP Proxy resolution (a possible evolution of slapd\-ldap(5)):"
662 In case the rewritten DN is an LDAP URI, the operation is initiated
663 towards the host[:port] indicated in the uri, if it does not refer
668 rewriteRule '^cn=root,.*' '%0' 'G{3}'
669 rewriteRule '^cn=[a-l].*' 'ldap://ldap1.my.org/%0' '@'
670 rewriteRule '^cn=[m-z].*' 'ldap://ldap2.my.org/%0' '@'
671 rewriteRule '.*' 'ldap://ldap3.my.org/%0' '@'
674 (Rule 1 is simply there to illustrate the `G{n}' action; it could have
678 rewriteRule '^cn=root,.*' 'ldap://ldap3.my.org/%0' '@'
681 with the advantage of saving one rewrite pass ...)
685 default slapd configuration file
692 Pierangelo Masarati, based on back-ldap by Howard Chu