| SLAPD-META(5) | File Formats Manual | SLAPD-META(5) |
slapd-meta - metadirectory backend to slapd
/etc/ldap/slapd.conf
The meta backend to slapd(8) performs basic LDAP proxying with respect to a set of remote LDAP servers, called "targets". The information contained in these servers can be presented as belonging to a single Directory Information Tree (DIT).
A basic knowledge of the functionality of the slapd-ldap(5) backend is recommended. This backend has been designed as an enhancement of the ldap backend. The two backends share many features (actually they also share portions of code). While the ldap backend is intended to proxy operations directed to a single server, the meta backend is mainly intended for proxying of multiple servers and possibly naming context masquerading. These features, although useful in many scenarios, may result in excessive overhead for some applications, so its use should be carefully considered. In the examples section, some typical scenarios will be discussed.
The proxy instance of slapd(8) must contain schema information for the attributes and objectClasses used in filters, request DN and request-related data in general. It should also contain schema information for the data returned by the proxied server. It is the responsibility of the proxy administrator to keep the schema of the proxy lined up with that of the proxied server.
Note: When looping back to the same instance of slapd(8), each connection requires a new thread; as a consequence, slapd(8) must be compiled with thread support, and the threads parameter may need some tuning; in those cases, unless the multiple target feature is required, one may consider using slapd-relay(5) instead, which performs the relayed operation internally and thus reuses the same connection.
There are examples in various places in this document, as well as in the slapd/back-meta/data/ directory in the OpenLDAP source tree.
These slapd.conf options apply to the META backend database. That is, they must follow a "database meta" line and come before any subsequent "backend" or "database" lines. Other database options are described in the slapd.conf(5) manual page.
Note: In early versions of back-ldap and back-meta it was recommended to always set
lastmod off
for ldap and meta databases. This was required because operational attributes related to entry creation and modification should not be proxied, as they could be mistakenly written to the target server(s), generating an error. The current implementation automatically sets lastmod to off, so its use is redundant and should be omitted.
Target configuration starts with the "uri" directive. All the configuration directives that are not specific to targets should be defined first for clarity, including those that are common to all backends. They are:
Target specification starts with a "uri" directive:
suffix "dc=foo,dc=com" uri "ldap://x.foo.com/dc=x,dc=foo,dc=com"
suffix "dc=foo,dc=com" uri "ldap://l1.foo.com/dc=foo,dc=com" "ldap://l2.foo.com/"
In a search request, if the search filter matches the pattern the target is considered while fulfilling the request; otherwise the target is ignored. There may be multiple occurrences of the filter directive for each target.
idassert-bind bindmethod=none|simple|sasl [binddn=<simple DN>] [credentials=<simple password>] [saslmech=<SASL mech>] [secprops=<properties>] [realm=<realm>] [authcId=<authentication ID>] [authzId=<authorization ID>] [authz={native|proxyauthz}] [mode=<mode>] [flags=<flags>] [starttls=no|yes|critical] [tls_cert=<file>] [tls_key=<file>] [tls_cacert=<file>] [tls_cacertdir=<path>] [tls_reqcert=never|allow|try|demand] [tls_cipher_suite=<ciphers>] [tls_protocol_min=<major>[.<minor>]] [tls_crlcheck=none|peer|all]
none|simple|sasl
where none is the default, i.e. no identity assertion is performed.
The authz parameter is used to instruct the SASL bind to exploit native SASL authorization, if available; since connections are cached, this should only be used when authorizing with a fixed identity (e.g. by means of the authzDN or authzID parameters). Otherwise, the default proxyauthz is used, i.e. the proxyAuthz control (Proxied Authorization, RFC 4370) is added to all operations.
The supported modes are:
<mode> := {legacy|anonymous|none|self}
If <mode> is not present, and authzId is given, the proxy always authorizes that identity. <authorization ID> can be
u:<user>
[dn:]<DN>
The former is supposed to be expanded by the remote server according to the authz rules; see slapd.conf(5) for details. In the latter case, whether or not the dn: prefix is present, the string must pass DN validation and normalization.
The default mode is legacy, which implies that the proxy will either perform a simple bind as the authcDN or a SASL bind as the authcID and assert the client's identity when it is not anonymous. Direct binds are always proxied. The other modes imply that the proxy will always either perform a simple bind as the authcDN or a SASL bind as the authcID, unless restricted by idassert-authzFrom rules (see below), in which case the operation will fail; eventually, it will assert some other identity according to <mode>. Other identity assertion modes are anonymous and self, which respectively mean that the empty or the client's identity will be asserted; none, which means that no proxyAuthz control will be used, so the authcDN or the authcID identity will be asserted. For all modes that require the use of the proxyAuthz control, on the remote server the proxy identity must have appropriate authzTo permissions, or the asserted identities must have appropriate authzFrom permissions. Note, however, that the ID assertion feature is mostly useful when the asserted identities do not exist on the remote server.
Flags can be
override,[non-]prescriptive,proxy-authz-[non-]critical
When the override flag is used, identity assertion takes place even when the database is authorizing for the identity of the client, i.e. after binding with the provided identity, and thus authenticating it, the proxy performs the identity assertion using the configured identity and authentication method.
When the prescriptive flag is used (the default), operations fail with inappropriateAuthentication for those identities whose assertion is not allowed by the idassert-authzFrom patterns. If the non-prescriptive flag is used, operations are performed anonymously for those identities whose assertion is not allowed by the idassert-authzFrom patterns.
When the proxy-authz-non-critical flag is used (the default), the proxyAuthz control is not marked as critical, in violation of RFC 4370. Use of proxy-authz-critical is recommended.
The TLS settings default to the same as the main slapd TLS settings, except for tls_reqcert which defaults to "demand".
The identity associated to this directive is also used for privileged operations whenever idassert-bind is defined and acl-bind is not. See acl-bind for details.
[<d>d][<h>h][<m>m][<s>[s]]
where <d>, <h>, <m> and <s> are respectively treated as days, hours, minutes and seconds. If set before any target specification, it affects all targets, unless overridden by any per-target directive.
<rule>: [dn[.<style>]:]<pattern>
<style>: subtree|children|regex
When <style> is either subtree or children the <pattern> is a DN that must be within the naming context served by the target. When <style> is regex the <pattern> is a regex(5) pattern. If the dn.<style>: prefix is omitted, dn.subtree: is implicitly assumed for backward compatibility.
In the subtree-exclude form if the request DN matches at least one rule, the target is not considered while fulfilling the request; otherwise, the target is considered based on the value of the request DN. When the request is a search, also the scope is considered.
In the subtree-include form if the request DN matches at least one rule, the target is considered while fulfilling the request; otherwise the target is ignored.
| match | exclude |
+---------+---------+-------------------+
| T | T | not candidate |
| F | T | continue checking |
+---------+---------+-------------------+
| T | F | candidate |
| F | F | not candidate |
+---------+---------+-------------------+
<op> ::= bind, add, delete, modrdn, modify, compare, search
The overall duration of the search operation is controlled either by the timelimit parameter or by server-side enforced time limits (see timelimit and limits in slapd.conf(5) for details). This timeout parameter controls how long the target can be irresponsive before the operation is aborted. Timeout is meaningless for the remaining operations, unbind and abandon, which do not imply any response, while it is not yet implemented in currently supported extended operations. If no operation is specified, the timeout val affects all supported operations. If specified before any target definition, it affects all targets unless overridden by per-target directives.
Note: if the timeout is exceeded, the operation is cancelled (according to the cancel directive); the protocol does not provide any means to rollback operations, so the client will not be notified about the result of the operation, which may eventually succeeded or not. In case the timeout is exceeded during a bind operation, the connection is destroyed, according to RFC4511.
A powerful (and in some sense dangerous) rewrite engine has been added to both the LDAP and Meta backends. While the former can gain limited beneficial effects from rewriting stuff, the latter can become an amazingly powerful tool.
Consider a couple of scenarios first.
1) Two directory servers share two levels of naming context; say "dc=a,dc=foo,dc=com" and "dc=b,dc=foo,dc=com". Then, an unambiguous Meta database can be configured as:
database meta suffix "dc=foo,dc=com" uri "ldap://a.foo.com/dc=a,dc=foo,dc=com" uri "ldap://b.foo.com/dc=b,dc=foo,dc=com"
Operations directed to a specific target can be easily resolved because there are no ambiguities. The only operation that may resolve to multiple targets is a search with base "dc=foo,dc=com" and scope at least "one", which results in spawning two searches to the targets.
2a) Two directory servers don't share any portion of naming context, but they'd present as a single DIT [Caveat: uniqueness of (massaged) entries among the two servers is assumed; integrity checks risk to incur in excessive overhead and have not been implemented]. Say we have "dc=bar,dc=org" and "o=Foo,c=US", and we'd like them to appear as branches of "dc=foo,dc=com", say "dc=a,dc=foo,dc=com" and "dc=b,dc=foo,dc=com". Then we need to configure our Meta backend as:
database meta suffix "dc=foo,dc=com" uri "ldap://a.bar.com/dc=a,dc=foo,dc=com" suffixmassage "dc=a,dc=foo,dc=com" "dc=bar,dc=org" uri "ldap://b.foo.com/dc=b,dc=foo,dc=com" suffixmassage "dc=b,dc=foo,dc=com" "o=Foo,c=US"
Again, operations can be resolved without ambiguity, although some rewriting is required. Notice that the virtual naming context of each target is a branch of the database's naming context; it is rewritten back and forth when operations are performed towards the target servers. What "back and forth" means will be clarified later.
When a search with base "dc=foo,dc=com" is attempted, if the scope is "base" it fails with "no such object"; in fact, the common root of the two targets (prior to massaging) does not exist. If the scope is "one", both targets are contacted with the base replaced by each target's base; the scope is derated to "base". In general, a scope "one" search is honored, and the scope is derated, only when the incoming base is at most one level lower of a target's naming context (prior to massaging).
Finally, if the scope is "sub" the incoming base is replaced by each target's unmassaged naming context, and the scope is not altered.
2b) Consider the above reported scenario with the two servers sharing the same naming context:
database meta suffix "dc=foo,dc=com" uri "ldap://a.bar.com/dc=foo,dc=com" suffixmassage "dc=foo,dc=com" "dc=bar,dc=org" uri "ldap://b.foo.com/dc=foo,dc=com" suffixmassage "dc=foo,dc=com" "o=Foo,c=US"
All the previous considerations hold, except that now there is no way to unambiguously resolve a DN. In this case, all the operations that require an unambiguous target selection will fail unless the DN is already cached or a default target has been set. Practical configurations may result as a combination of all the above scenarios.
Note on ACLs: at present you may add whatever ACL rule you desire to the Meta (and LDAP) backends. However, the meaning of an ACL on a proxy may require some considerations. Two philosophies may be considered:
a) the remote server dictates the permissions; the proxy simply passes back what it gets from the remote server.
b) the remote server unveils "everything"; the proxy is responsible for protecting data from unauthorized access.
Of course the latter sounds unreasonable, but it is not. It is possible to imagine scenarios in which a remote host discloses data that can be considered "public" inside an intranet, and a proxy that connects it to the internet may impose additional constraints. To this purpose, the proxy should be able to comply with all the ACL matching criteria that the server supports. This has been achieved with regard to all the criteria supported by slapd except a special subtle case (please file an ITS if you can find other exceptions: <http://www.openldap.org/its/>). The rule
access to dn="<dn>" attrs=<attr>
by dnattr=<dnattr> read
by * none
cannot be matched iff the attribute that is being requested, <attr>, is NOT <dnattr>, and the attribute that determines membership, <dnattr>, has not been requested (e.g. in a search)
In fact this ACL is resolved by slapd using the portion of entry it retrieved from the remote server without requiring any further intervention of the backend, so, if the <dnattr> attribute has not been fetched, the match cannot be assessed because the attribute is not present, not because no value matches the requirement!
Note on ACLs and attribute mapping: ACLs are applied to the mapped attributes; for instance, if the attribute locally known as "foo" is mapped to "bar" on a remote server, then local ACLs apply to attribute "foo" and are totally unaware of its remote name. The remote server will check permissions for "bar", and the local server will possibly enforce additional restrictions to "foo".
A string is rewritten according to a set of rules, called a `rewrite context'. The rules are based on POSIX (''extended'') regular expressions (regex) with substring matching; basic variable substitution and map resolution of substrings is allowed by specific mechanisms detailed in the following. The behavior of pattern matching/substitution can be altered by a set of flags.
The underlying concept is to build a lightweight rewrite module for the slapd server (initially dedicated to the LDAP backend).
An incoming string is matched against a set of rules. Rules are made of a regex match pattern, a substitution pattern and a set of actions, described by a set of flags. In case of match a string rewriting is performed according to the substitution pattern that allows one to refer to substrings matched in the incoming string. The actions, if any, are finally performed. The substitution pattern allows map resolution of substrings. A map is a generic object that maps a substitution pattern to a value. The flags are divided in "Pattern matching Flags" and "Action Flags"; the former alter the regex match pattern behavior while the latter alter the action that is taken after substitution.
The ordering of the flags can be significant. For instance: `IG{2}' means ignore errors and jump two lines ahead both in case of match and in case of error, while `G{2}I' means ignore errors, but jump two lines ahead only in case of match.
More flags (mainly Action Flags) will be added as needed.
See regex(7) and/or re_format(7).
Everything starting with `%' requires substitution;
the only obvious exception is `%%', which is left as is;
the basic substitution is `%d', where `d' is a digit; 0 means the whole string, while 1-9 is a submatch;
a `%' followed by a `{' invokes an advanced substitution. The pattern is:
where <name> must be a legal name for the map, i.e.
<name> ::= [a-z][a-z0-9]* (case insensitive) <op> ::= `>' `|' `&' `&&' `*' `**' `$'
and <substitution> must be a legal substitution pattern, with no limits on the nesting level.
The operators are:
Substitution escaping has been delegated to the `%' symbol, which is used instead of `\' in string substitution patterns because `\' is already escaped by slapd's low level parsing routines; as a consequence, regex escaping requires two `\' symbols, e.g. `.*\.foo\.bar' must be written as `.*\\.foo\\.bar'.
A rewrite context is a set of rules which are applied in sequence. The basic idea is to have an application initialize a rewrite engine (think of Apache's mod_rewrite ...) with a set of rewrite contexts; when string rewriting is required, one invokes the appropriate rewrite context with the input string and obtains the newly rewritten one if no errors occur.
Each basic server operation is associated to a rewrite context; they are divided in two main groups: client -> server and server -> client rewriting.
client -> server:
(default) if defined and no specific context
is available bindDN bind searchBase search searchFilter search searchFilterAttrDN search compareDN compare compareAttrDN compare AVA addDN add addAttrDN add AVA modifyDN modify modifyAttrDN modify AVA modrDN modrdn newSuperiorDN modrdn deleteDN delete exopPasswdDN password modify extended operation DN if proxy
server -> client:
searchResult search (only if defined; no default;
acts on DN and DN-syntax attributes
of search results) searchAttrDN search AVA matchedDN all ops (only if applicable)
# set to `off' to disable rewriting rewriteEngine on # the rules the "suffixmassage" directive implies rewriteEngine on # all dataflow from client to server referring to DNs rewriteContext default rewriteRule "(.*)<virtualnamingcontext>$" "%1<realnamingcontext>" ":" # empty filter rule rewriteContext searchFilter # all dataflow from server to client rewriteContext searchResult rewriteRule "(.*)<realnamingcontext>$" "%1<virtualnamingcontext>" ":" rewriteContext searchAttrDN alias searchResult rewriteContext matchedDN alias searchResult # Everything defined here goes into the `default' context. # This rule changes the naming context of anything sent # to `dc=home,dc=net' to `dc=OpenLDAP, dc=org' rewriteRule "(.*)dc=home,[ ]?dc=net"
"%1dc=OpenLDAP, dc=org" ":" # since a pretty/normalized DN does not include spaces # after rdn separators, e.g. `,', this rule suffices: rewriteRule "(.*)dc=home,dc=net"
"%1dc=OpenLDAP,dc=org" ":" # Start a new context (ends input of the previous one). # This rule adds blanks between DN parts if not present. rewriteContext addBlanks rewriteRule "(.*),([^ ].*)" "%1, %2" # This one eats blanks rewriteContext eatBlanks rewriteRule "(.*),[ ](.*)" "%1,%2" # Here control goes back to the default rewrite # context; rules are appended to the existing ones. # anything that gets here is piped into rule `addBlanks' rewriteContext default rewriteRule ".*" "%{>addBlanks(%0)}" ":" # Rewrite the search base according to `default' rules. rewriteContext searchBase alias default # Search results with OpenLDAP DN are rewritten back with # `dc=home,dc=net' naming context, with spaces eaten. rewriteContext searchResult rewriteRule "(.*[^ ]?)[ ]?dc=OpenLDAP,[ ]?dc=org"
"%{>eatBlanks(%1)}dc=home,dc=net" ":" # Bind with email instead of full DN: we first need # an ldap map that turns attributes into a DN (the # argument used when invoking the map is appended to # the URI and acts as the filter portion) rewriteMap ldap attr2dn "ldap://host/dc=my,dc=org?dn?sub" # Then we need to detect DN made up of a single email, # e.g. `mail=someone@example.com'; note that the rule # in case of match stops rewriting; in case of error, # it is ignored. In case we are mapping virtual # to real naming contexts, we also need to rewrite # regular DNs, because the definition of a bindDn # rewrite context overrides the default definition. rewriteContext bindDN rewriteRule "^mail=[^,]+@[^,]+$" "%{attr2dn(%0)}" ":@I" # This is a rather sophisticated example. It massages a # search filter in case who performs the search has # administrative privileges. First we need to keep # track of the bind DN of the incoming request, which is # stored in a variable called `binddn' with session scope, # and left in place to allow regular binding: rewriteContext bindDN rewriteRule ".+" "%{&&binddn(%0)}%0" ":" # A search filter containing `uid=' is rewritten only # if an appropriate DN is bound. # To do this, in the first rule the bound DN is # dereferenced, while the filter is decomposed in a # prefix, in the value of the `uid=<arg>' AVA, and # in a suffix. A tag `<>' is appended to the DN. # If the DN refers to an entry in the `ou=admin' subtree, # the filter is rewritten OR-ing the `uid=<arg>' with # `cn=<arg>'; otherwise it is left as is. This could be # useful, for instance, to allow apache's auth_ldap-1.4 # module to authenticate users with both `uid' and # `cn', but only if the request comes from a possible # `cn=Web auth,ou=admin,dc=home,dc=net' user. rewriteContext searchFilter rewriteRule "(.*\\()uid=([a-z0-9_]+)(\\).*)"
"%{**binddn}<>%{&prefix(%1)}%{&arg(%2)}%{&suffix(%3)}"
":I" rewriteRule "[^,]+,ou=admin,dc=home,dc=net"
"%{*prefix}|(uid=%{*arg})(cn=%{*arg})%{*suffix}" ":@I" rewriteRule ".*<>" "%{*prefix}uid=%{*arg}%{*suffix}" ":" # This example shows how to strip unwanted DN-valued # attribute values from a search result; the first rule # matches DN values below "ou=People,dc=example,dc=com"; # in case of match the rewriting exits successfully. # The second rule matches everything else and causes # the value to be rejected. rewriteContext searchResult rewriteRule ".*,ou=People,dc=example,dc=com" "%0" ":@" rewriteRule ".*" "" "#"
In case the rewritten DN is an LDAP URI, the operation is initiated towards the host[:port] indicated in the uri, if it does not refer to the local server. E.g.:
rewriteRule '^cn=root,.*' '%0' 'G{3}'
rewriteRule '^cn=[a-l].*' 'ldap://ldap1.my.org/%0' ':@'
rewriteRule '^cn=[m-z].*' 'ldap://ldap2.my.org/%0' ':@'
rewriteRule '.*' 'ldap://ldap3.my.org/%0' ':@'
(Rule 1 is simply there to illustrate the `G{n}' action; it could have been written:
rewriteRule '^cn=root,.*' 'ldap://ldap3.my.org/%0' ':@'
with the advantage of saving one rewrite pass ...)
The meta backend does not honor all ACL semantics as described in slapd.access(5). In general, access checking is delegated to the remote server(s). Only read (=r) access to the entry pseudo-attribute and to the other attribute values of the entries returned by the search operation is honored, which is performed by the frontend.
The proxy cache overlay allows caching of LDAP search requests (queries) in a local database. See slapo-pcache(5) for details.
The following statements have been deprecated and should no longer be used.
slapd.conf(5), slapd-ldap(5), slapo-pcache(5), slapd(8), regex(7), re_format(7).
Pierangelo Masarati, based on back-ldap by Howard Chu
| 2018/12/19 | OpenLDAP |