slapd-relay — relay backend to slapd
ETCDIR/slapd.conf
The primary purpose of this slapd(8) backend is to map
a naming context defined in a database running in the same
slapd(8) instance into a
virtual naming context, with attributeType and objectClass
manipulation, if required. It requires the rwm
overlay
.
This backend and the above mentioned overlay are experimental.
The following slapd.conf
directives apply
to the relay backend database. That is, they must follow a
"database relay" line and come before any subsequent
"backend" or "database" lines. Other database options are
described in the slapd.conf(5) manual page;
only the suffix
directive is required by the relay
backend.
The naming context of the database that is presented
under a virtual naming context. The presence of this
directive implies that one specific database, i.e. the
one serving the real naming
context, will be presented under a virtual
naming context. This directive automatically
instantiates the rwm
overlay. If the optional massage
keyword is
present, the suffix massaging is automatically
configured as well; otherwise, specific massaging
instructions are required by means of the rewrite
directives
described in slapo-rwm(5).
One important issue is that access rules are based on the
identity that issued the operation. After massaging from the
virtual to the real naming context, the frontend sees the
operation as performed by the identity in the real naming
context. Moreover, since back-relay
bypasses the real
database frontend operations by short-circuiting operations
thru the internal backend API, the original database access
rules do not apply but in selected cases, i.e. when the
backend itself applies access control. As a consequence, the
instances of the relay database must provide own access rules
that are consistent with those of the original database,
possibly adding further specific restrictions. So, access
rules in the relay
database must refer to identities in the real naming context.
Examples are reported in the EXAMPLES section.
If no relay
directive is given, the relay
database does not refer
to any specific database, but the most appropriate one is
looked-up after rewriting the request DN for the operation
that is being handled.
This allows to write carefully crafted rewrite rules that cause some of the requests to be directed to one database, and some to another; e.g., authentication can be mapped to one database, and searches to another, or different target databases can be selected based on the DN of the request, and so.
Another possibility is to map the same operation to different databases based on details of the virtual naming context, e.g. groups on one database and persons on another.
To implement a plain virtual naming context mapping that refers to a single database, use
database relay suffix "dc=virtual,dc=naming,dc=context" relay "dc=real,dc=naming,dc=context" massage
To implement a plain virtual naming context mapping that looks up the real naming context for each operation, use
database relay suffix "dc=virtual,dc=naming,dc=context" overlay rwm suffixmassage "dc=real,dc=naming,dc=context"
This is useful, for instance, to relay different databases that share the terminal portion of the naming context (the one that is rewritten).
To implement the old-fashioned suffixalias, e.g. mapping the virtual to the real naming context, but not the results back from the real to the virtual naming context, use
database relay suffix "dc=virtual,dc=naming,dc=context" relay "dc=real,dc=naming,dc=context" rewriteEngine on rewriteContext default rewriteRule "dc=virtual,dc=naming,dc=context" "dc=real,dc=naming,dc=context" ":@" rewriteContext searchFilter rewriteContext searchEntryDN rewriteContext searchAttrDN rewriteContext matchedDN
Note that the virtual database is bound to a single real database, so the rwm overlay is automatically instantiated, but the rewrite rules are written explicitly to map all the virtual to real naming context data flow, but none of the real to virtual.
Access rules:
database bdb suffix "dc=example,dc=com" # skip... access to dn.subtree="dc=example,dc=com" by dn.exact="cn=Supervisor,dc=example,dc=com" write by * read database relay suffix "o=Example,c=US" relay "dc=example,dc=com" massage # skip ... access to dn.subtree="o=Example,c=US" by dn.exact="cn=Supervisor,dc=example,dc=com" write by dn.exact="cn=Relay Supervisor,dc=example,dc=com" write by * read
Note that, in both databases, the identities (the
<who>
clause)
are in the real naming
context, i.e. `dc=example,dc=com'
, while
the targets (the <what>
clause) are in
the real
and in the
virtual naming
context, respectively.
The relay
backend does not honor any of the access control semantics
described in slapd.access(5); all access
control is delegated to the relayed database(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.