LDAP Tips and Tricks

Last modified 17 May 2024 08:42 +02:00

Table of Contents

LDAP in midPoint
LDAP Utilities
See Also

LDAP in midPoint

Mappings and DN

The simple way how to map common midPoint property to DN is this:

'cn=' + name + ',ou=people,dc=example,dc=com'

That’s fine for some cases. But if the name may contain commas, non-ascii characters or similar nasties then the operations will break because the result is not a valid DN.

The more reliable way to construct DN is to use JNDI classes:

import javax.naming.ldap.Rdn
import javax.naming.ldap.LdapName

dn = new LdapName('ou=people,dc=example,dc=com')
dn.add(new Rdn('cn',name.toString()))
return dn.toString()

This works quite reliably, but it is not very elegant. Therefore tow new Script Expression Functions were added in midPoint 3.4 and 3.3.1: composeDn() and composeDnWithSuffix():

basic.composeDnWithSuffix('cn', name, 'ou=people,dc=example,dc=com')

This is reliable, simple and elegant.

DistinguishedName Matching Rule

The DN is a bit tricky thing. The following DNs are all equivalent:

CN=foo,OU=people,DC=example,DC=com
cn=foo,ou=People,dc=example,dc=com
cn=foo, ou=People, dc=example, dc=com
cn=foo, ou=people, dc=example,dc=com

It is quite common that the LDAP server returns the DN in a different form that was used to create the object. Therefore object created with “cn=foo, ou=People, dc=example, dc=com” will return from search with a DN of “CN=foo,OU=people,DC=example,DC=com”. MidPoint is very conservative by default and it assumes that all identifiers match exactly (case, whitespaces and all). Therefore if the server returns a different form of DN then it means trouble in the default configuration. The most likely symptom is that the shadows that represent the LDAP objects will duplicate. This is understandable given the default conservative identifier handling as, strictly speaking, these are two different identifiers and therefore midPoint assumes that two different objects are identified by them.

Fortunately, midPoint is designed to handle situations such as this by using matching rules. And there is a very convenient matching rule that understands the DN syntax: distinguishedName. All that is needed is a proper configuration of a matching rules for all the attributes that may contain DN:

            <attribute>
                <ref>ri:dn</ref>
                ...
                <matchingRule>mr:distinguishedName</matchingRule>
                ...
            </attribute>

That is all. Now midPoint knows how to reliably compare DN values and it takes care of the rest.

member, memberOf and others

 Do not forget that DNs can also appear in other attributes, not just the object identifier.
The most common case are group membership attributes such as `member`, `memberOf` and similar.
Those also need matching rules for the system to operate properly.

Since midPoint 3.3 the matching rules are determined automatically - assuming that the directory server properly advertises the data types in the schema. If you work with a good LDAP server then there is no need to explicitly set up matching rules.

Base Context

LDAP objects may be stored in several subtrees of the directory information tree. E.g. one type of groups may be in ou=groups,dc=example,dc=com, a different type may be in ou=projects,dc=example,dc=com. These groups are segregated into subtrees because they may have the same names (cn), or they may have different access control and so on. However these groups may have the same objectclass and overall a very similar structure. As they have the same object class then they are normally indistinguishable by midPoint (remember: midPoint is generic, it does not understand the structure of a DN). But there is a way how to distinguish them by using the base context feature:

<resource>
    ...
    <schemaHandling>
        ...
        <objectType>
            <kind>entitlement</kind>
            <intent>project</intent>
            <objectClass>ri:groupOfNames</objectClass>
            <baseContext>
                <objectClass>ri:organizationalUnit</objectClass>
                <filter>
                    <q:text>attributes/dn = "ou=projects,dc=example,dc=com"</q:text>
                </filter>
            </baseContext>
            ...

LDAP Utilities

All of this assumes OpenLDAP set of LDAP utilities (this is the standard on most Linux boxes):

VLV LDAP Search

ldapsearch -h localhost -D uid=idm,ou=Administrators,dc=example,dc=com -w secret -b dc=example,dc=com -E \!sss=uid:2.5.13.3 -E \!vlv=0/9/1/0

This returns the first ten entries under the suffix as ordered by entryUUID.

The 0/9/1/0 means return zero entries before "target", return 9 entries after "target" (which means 10 entries), the first entry has index 1 and we do have no idea how many entries are the in total (the last zero).

Hash Password

slappasswd -s secret

Root DSE

ldapsearch -x -h localhost -b "" -s base "(objectclass=*)" * +

Disabling Server Certificate validation

/etc/ldap/ldap.conf

TLS_REQCERT    allow