6 Configuration of Directory Synchronization Profiles

The SearchDeltaSize Parameter

The SearchDeltaSize parameter determines how many incremental changes are processed during each iteration in a synchronization cycle. By default, the SearchDeltaSize parameter is assigned a value of 500. In some cases, you will experience better synchronization efficiency if you assign a higher value to this parameter. However, be sure that the value you specify does not exceed the LDAP search limit of the connected directory server. Otherwise, you may receive an error during synchronization and some changes may not be processed.

The SkipErrorToSyncNextChange Parameter

The SkipErrorToSyncNextChange parameter determines how the Oracle directory integration and provisioning server handles an error when processing a change during synchronization. By default, the SkipErrorToSyncNextChange parameter is assigned a value of false, which means that the Oracle directory integration and provisioning server will continue processing a change until the error is resolved. If you assign a value of true to the SkipErrorToSyncNextChange parameter, the Oracle directory integration and provisioning server will skip any changes that cause an error. Any failures are recorded in the $ORACLE_HOME/ldap/odi/log/profilename.aud audit log file. If you do assign a value of true to the SkipErrorToSyncNextChange parameter, be sure to periodically review the audit log for failures.

Distinguished Name Mapping

This section specifies how entries are mapped between Oracle Internet Directory and a connected directory. If the mapping is between Oracle Internet Directory and another LDAP directory, then you can create multiple mapping rules, as explained in "Configuring Mapping Rules". The domain rule specifications appear after a line containing only the keyword DomainRules. Each domain rule is represented with the components, separated by colons, that are described in Table 6-2.

Attribute-Level Mapping

The attribute rule specifications appear after a line containing only the keyword AttributeRules. Attribute rules specify how property values for an entry are related between two LDAP directories. For example, the cn attribute of a user object in one directory can be mapped to the givenname object in another directory. Similarly, the cn attribute of a group object in one directory can be mapped to the displayname attribute in another directory. Each attribute rule is represented with the components, separated by colons, and described in Table 6-3. The attribute rule specifications end with a line containing only the characters ###.

In a newly created synchronization profile, mapping rules are empty. To enter mapping rules, edit a file that strictly follows the correct format.

How to Construct a New Mapping File

To create a new mapping file, follow these steps:

1. Identify the container(s) of interest for synchronization in the source directory.

2. Identify the destination container or containers to which the objects in the source containers should be mapped to. Be sure that the specified container already exists in the directory.

3. Determine the rule to create a DN of the entry to be created in the destination directory. In LDAP-to-LDAP, mapping is normally one-to-one. In non-LDAP-to-LDAP, a domain, DN construct rule is required. For instance in the case of synchronizing from a tagged file or Human Resources agent, the mapping rule may be of the form uid=%,dc=mycompany,dc=com. In this case, the uid attribute must be present in all the changes to be applied from Oracle Human Resources. The uid attribute must be specified as a required attribute, as specified in step 6.

4. Identify the objects that you want to synchronize between directories—that is, the relevant object classes in the source and destination directories. In general, objects that get synchronized between directories include users, groups, organizational units, organizations, and other resources. Identify the actual object classes used in the directories to identify these objects.

5. Identify the properties of the various objects that you want to synchronize between directories—that is, the attributes in the LDAP context. All the attributes of an object need not be synchronized. The properties of users that you might want to synchronize are cn, sn, uid, mail.

6. Define the mapping rules. Each mapping rule has this format:

   <srcAttrName>:<ReqdFlag>:<srcAttrType>:<SrcObjectClass>: <dstAttrName>:<dstAttrType>:<dstObjectClass>: <Mapping Rule>
   

   While defining the mapping rule, ensure the following:

   • Every required attribute has a sequence number. For example, if in step 3 the uid attribute is identified as required, then assign a value of 1 in place of <ReqdFlag>.

   • Every relevant object class has a schema definition on the destination directory.

   • Every mandatory attribute in a destination object class has a value assigned from the source. This holds good even for standard object classes also, as the different LDAP implementations may not be completely standards-compliant.

   It is not necessary to assign all attributes belonging to a source object class to a single destination object class. Different attributes of a source object class can be assigned to different attributes belonging to different destination object classes.

   If an attribute has binary values, then specify it as binary in the <attrtype> field.

Mapping rules are flexible: They can include both one-to-many and many-to-one mappings.

• One-to-many

  One attribute in a connected directory can map to many attributes in Oracle Internet Directory. For example, suppose an attribute in the connected directory is Address:123 Main Street/MyTown, MyState 12345. You can map this attribute in Oracle Internet Directory to both the LDAP attribute homeAddress and the LDAP attribute postalAddress.

• Many-to-one

  Multiple attributes in a connected directory can map to one attribute in Oracle Internet Directory. For example, suppose that the Oracle Human Resources directory represents Anne Smith by using two attributes: firstname=Anne and lastname=Smith. You can map these two attributes to one attribute in Oracle Internet Directory: cn=Anne Smith. However, in bidirectional synchronization, you cannot then map in reverse. For example, you cannot map cn=Anne Smith to many attributes.

Supported Attribute Mapping Rules and Examples

The attribute mapping rules supported are:

• Concatenation (+): Used to concatenate two string attributes

  The mapping rule looks like:

  Firstname,lastname:  :  :  : givenname:  : inetorgperson: firstname+lastname
  
  

  For example, if the Firstname is John and LastName is Doe in the source, then this rule results in the givenname attribute in the destination with the value JohnDoe.

• OR operator ( | ): Used to assign one of the values of the two string attributes to the destination

  The mapping rule looks like this:

  Fistname,lastname : : : :givenname: :inetorgperson: firstname | lastname
  
  

  In this example, givenname is assigned the value of firstname if it exists. If the firstname attribute does not exist, then givenname is assigned the value of lastname. If both the values are empty, then no value is assigned.

• bin2b64 ( ): Used to store a binary value of the source directory as a base64 encoded value in the destination directory. Typical usage is as follows:

  objectguid: : : :binary: :orclobjectguid: orcladuser:bin2b64(objectguid)
  
  

  This is required when you need search on the value of (objectguid).

• tolower(): Convert the String attribute value to lowercase.

  firstname: : : :givenname: :inetorgperson: tolower(firstname)
  
  

• toupper (): Convert the String attribute value to uppercase.

  firstname: : : :givenname: :inetorgperson: toupper(firstname)
  
  

• trunc(str,char): Truncate the string beginning from the first occurrence of the specified char

  mail : : : : uid : : inetorgperson : trunc(mail,'@')
  
  

  For example, if mail is John.Doe@acme.com in the source, then this rule results in the uid attribute in the destination with the value "John.Doe"

• truncl(str, char): Truncate the string up to and including the first occurrence of the specified char

  mail : : : : uid : : inetorgperson : truncl(mail,'@')
  
  

  For example, if mail is John.Doe@acme.com in the source, then this rule results in the uid attribute in the destination with the value acme.com.

• trunc(str1, str2): Truncate the string beginning with the first occurrence of the specified string

  mail : : : : uid : : inetorgperson : truncl(mail, "@")
  
  

• dnconvert (str): Used for DN type attributes if domain mapping is used.

  This example assumes the following domain mapping rule:

  DomainRules
  cn=srcdomain:cn=dstdomain:
  
  

  For example:

  uniquemember : : : groupofuniquenames : uniquemember : :groupofuniquenames : dnconvert(uniquemember)
  
  

  In this example, if uniquemember in the source is cn=testuser1,cn=srcdomain, then uniquemember in the destination becomes cn=test user1,cn=dstdomain.

• Literals:

  Userpassword: : :person: userpassword: :person: 'welcome1'
  

Example: A Mapping File for a TAGGED-File Interface

Based on the preceding discussions, here is a sample mapping file for importing user entries from the Oracle Human Resources database tables by using the tagged-file interface. Note that the source is a non-LDAP directory. This sample file is supplied during installation, at $ORACLE_HOME/ldap/odi/conf/oraclehragent.map.master.

DomainRules
NONLDAP:dc=myCompany,dc=com:uid=%dc=myCompany,dc=com
AttributeRules
firstname: : : :cn: :person
email : : : :cn: :person: trunc(email,'@')
email : 1 : :uid: :person:trunc(email,'@')
firstname,lastname: : : :cn: :person: firstname+","+lastname
lastname,firstname: : : :cn: :person: lastname+","+firstname
firstname,lastname: : : :sn: :person: lastname | firstname
EmployeeNumber: : : :employeenumber: :inetOrgperson
EMail: : : :mail: :inetOrgperson
TelephoneNumber1: : : :telephonenumber: :person
TelephoneNumber2: : : :telephonenumber: :person
TelephoneNumber3: : : :telephonenumber: :person
Address1: : : :postaladdress: :person
state: : : :st: :locality
street1: : : :street: :locality
zip: : : :postalcode: :locality
town_or_city: : : :l: :locality
Title: : : :title: :organizationalperson
#Sex: : : :sex: :person ###

As described earlier, the mapping file consists of keywords and a set of domain and attribute mapping rule entries. The mapping file in this example contains the domain rule NONLDAP:dc=myCompany,dc=com:cn=%,dc=myCompany,dc=com.

• This rule implies that the source domain is NONLDAP—that is, there is no source domain.

• The destination domain (:dc=myCompany,dc=com) implies that all the directory entries this profile deals with are in the domain dc=myCompany,dc=com. Be sure that the domain exists before the start of synchronization.

• The domain mapping rule (:uid=%,dc=myCompany,dc=com) implies that the data from the source should refer to the entry in the directory with the DN that is constructed using this domain mapping rule. In this case, uid must be one of the destination attributes that should always have a non-null value. If any data corresponding to an entry to be synchronized has a null value, then the mapping engine assumes that the entry is invalid and proceeds to the next entry. To identify the entry correctly in the directory, it is also necessary that uid should be single-valued.

• In the case of the tagged file, the source entry does not have any object class to indicate the type of object it is synchronizing. Note that the SrcObjectClass field is empty.

• Every object whose destination is Oracle Internet Directory must have an object class. Specify an object class for every attribute.

• Note that email is specified as a required attribute in the sample mapping file. This is because the uid attribute is derived from the email attribute. Successful synchronization requires the email attribute to be specified in all changes specified in the tagged file as follows:

  Email : 1 : : :uid : : person : trunc(email,'@')
  
  

• In some cases, the RDN of the DN needs to be constructed by using the name of a multivalued attribute. For example, to construct an entry with the DN of cn=%,l=%,dc=myCompany,dc=com, where cn is a multivalued attribute, the DomainMappingRule can be of this form: rdn,l=%,dc=myCompany,dc=com where rdn is one of the destination attributes having a non-null value. A typical mapping file supporting this could have the following form:

  DomainRules
  NONLDAP:dc=us,dc=myCompany,dc=com:rdn,l=%,dc=us,dc=myCompany,dc=com
  AttributeRules
  firstname: : :cn: :person
  email : : : :cn: :person: trunc(email,'@')
  email : 1: : :rdn: :person: 'cn='+trunc(email,'@')
  firstname,lastname: : : :cn: :person: firstname+","+lastname
  lastname,firstname: : : :cn: :person: lastname+","+firstname
  firstname,lastname: : : :sn: :person: lastname | firstname
  EmployeeNumber: : : :employeenumber: :inetOrgperson
  EMail: : : :mail: :inetOrgperson
  TelephoneNumber1: : : :telephonenumber: :person
  TelephoneNumber2: : : :telephonenumber: :person
  TelephoneNumber3: : : :telephonenumber: :person
  Address1: : : :postaladdress: :person
  Address1: : : :postaladdress: :person
  Address1: : : :postaladdress: :person
  state: : : :st: :locality
  street1: : : :street: :locality
  zip: : : :postalcode: :locality
  town_or_city: 2 : : :1: :locality
  Title: : : :title: :organizationalperson
  #Sex: : : :sex: :person
  ###
  
  

Example: Mapping Files for an LDIF Interface

A set of sample integration profiles are created as part of installation by using the Directory Integration and Provisioning Assistant. The properties file used for creating the profile is located in the directory $ORACLE_HOME/ldap/odi/samples.

Sample Import Mapping File

DomainRules
dc=mycompany.oid,dc=com:dc=mycompany.iplanet,dc=com
AttributeRules
# Mapping rules to map the domains and containers
o: : :organization: o: :organization
ou: : :organizationalUnit: ou: : organizationalUnit
dc: : :domain:dc: :domain
# Mapping Rules to map users
uid : : :person: uid: :inetOrgperson
sn: : :person:sn: :person
cn: : :person:cn: :person
mail: :inetorgperson: mail: :inetorgperson
employeenumber: :organizationalPerson: employeenumber: :organizationalperson
c: : :country:c: :country
l: : :locality: l: :locality
telephonenumber: :organizationalPerson: telephonenumber: :organizationalperson
userpassword: : :person: userpassword: :person
uid: : :person: orcldefaultProfileGroup: :orclUserV2
# Mapping Rules to map groups
cn: : :groupofuniquenames:cn: :groupofuniquenames
member: : :groupofuniquenames:member: :orclgroup
uniquemember: : :groupofuniquenames:uniquemember: :orclgroup
owner: : :groupofuniquenames:owner: :orclgroup
# userpassword: :base64:userpassword: :binary:
 

Notice in the preceding example that both the source domain and destination domain are specified in the Domain Mapping rule section. In this example, the source and the destination domains are the same. However, you can specify a different destination domain, provided the container exists in the destination directory.

Also notice in the preceding example that the attribute rules are divided into two sections: user attribute mapping rules and group attribute mapping rules. Specifying the object class in a mapping rule helps to uniquely map a specific attribute of an object.

Updating Mapping Rules

You can customize mapping rules by adding new ones, modifying existing ones, or deleting some from the mapping rule set specified in the orclodipAttributeMappingRules attribute. In general, to perform any of these operations, you identify the file containing the mapping rules, or store the value of the attribute for a file by using an ldapsearch command as described in the Oracle Internet Directory Administrator's Guide.

You cannot edit the mapping rules in the Oracle Directory Integration and Provisioning Server Administration tool. Instead, mapping rules are stored in a file that you upload to the directory as a value of the attribute. To upload the mapping file, use the Directory Integration and Provisioning Assistant. Once you have created and uploaded the mapping file, you can maintain a copy of it in the $ORACLE_HOME/ldap/odi/conf directory, and upload it again after any future update.

dipassistant mp -profile profile name odip.profile.mapfile=map file

Filtering Changes with an LDAP Search

For connected directories that obtain incremental changes through an LDAP search, such as Active Directory, use the following syntax to assign a value to the searchfilter attribute of either the Connected Directory Matching Filter (orclODIPConDirMatchingFilter) or the OID Matching Filter (orclODIPOIDMatchingFilter):

"searchfilter=LDAP_SEARCH_FILTER"
 

The following example creates an LDAP search filter that retrieves organizational units, groups, and users, but not computers:

"searchfilter=(|(objectclass=group)(objectclass=organizationalUnit)
(&(objectclass=user)(!(objectclass=computer))))"

Filtering Changes from a Change Log

For connected directories that store changes in a change log, you can use just the following simple operators, which are provided by Oracle Directory Integration and Provisioning, to specify a matching filter for either the Connected Directory Matching Filter (orclODIPConDirMatchingFilter) or the OID Matching Filter (orclODIPOIDMatchingFilter):

• = (equal operator)

• != (not equal operator)

You can use the preceding operators with either LDAP or non-LDAP directories, provided they obtain incremental changes from a change log. Wildcards and pattern matching are not supported with the preceding operators if you do not use the searchfilter attribute. However, when multiple operator pairs are including in the filter, the expression is evaluated as a logical AND operation. For example, the following expression includes four operator pairs:

"(objectclass=group)(objectclass=organizationalUnit)
(objectclass=user)(objectclass!=computer)"
 

The preceding expression evaluates as follows:

objectclass is equal to group
AND objectclass is equal to organizationalUnit
AND objectclass is equal to user
AND objectclass is NOT equal to computer
 

For connected directories that store changes in a change log, a matching filter can synchronize changes for only the attributes that appear in the change log. If you include attributes in a matching filter that do not appear in the change log, the search operation will fail. For this reason, matching filters are of limited use for connected directories that store incremental changes in a change log.