This document is a work in progress. |
Real-time provisioning is the incremental provisioning of groups, stems, and memberships triggered from the Grouper change log. Incremental provisioning is distinguished from full provisioning in that only a single or subset of an attribute's values are provisioned.
Real-time provisioning is available from the provisioning service provider (psp, formerly known as ldappcng) as of version 2.1.0. It should be possible to install and configure the psp for Grouper versions 1.6 and up.
These instructions assume that Grouper subjects are already provisioned to your ldap directory, and makes use of the vt-ldap based ldap source adapter.
These instructions were written on a Mac, other platforms should be similar.
Install the provisioning service provider by copying jar files and example configuration files from the psp distribution to your Grouper API installation.
Many jars are dependencies of the Shibboleth attribute resolver and may not be necessary in your deployment.
To install Grouper including the API, UI, WS, grouperClient, psp, etc., download and run the Grouper Installer.
As of version 2.1.0, the psp may be installed but not configured, via the Grouper Installer.
curl http://www.internet2.edu/grouper/release/2.1.0/grouperInstaller.jar -O java -jar grouperInstaller.jar |
To install manually, download and unpack the psp, then copy jars and configuration files to your Grouper API installation.
Download the PSP from ...
The distribution name is of the form grouper.psp-${version}.tar.gz
curl http://www.internet2.edu/grouper/release/2.1.0/grouper.psp-2.1.0.tar.gz -O tar xzf grouper.psp-2.1.0.tar.gz |
Copy jars located in lib/custom
from the psp distribution to the Grouper API installation.
cp -vR grouper.psp-2.1.0-SNAPSHOT/lib/custom/ grouper.apiBinary-2.1.0/lib/custom/ |
Copy example configuration files located in conf
from the psp distribution to the Grouper API installation.
cp -vR grouper.psp-2.1.0/conf/ grouper.apiBinary-2.1.0/conf/ |
Example psp configuration files should have been installed to the conf
directory of the Grouper API installation.
Example psp configuration files are in directories named psp-example-*
.
The Grouper to LDAP configuration examples assume that Grouper subjects exist in an ldap directory.
This configuration example should apply to any ldap directory server.
This configuration example targets an IBM Tivoli Directory Server with requirements from Penn State.
examples |
|
DN structure |
flat |
|
member subject ids |
|
group identifiers |
|
member names |
|
group names |
This configuration example applies to OpenLDAP without the memberOf overlay slapo-memberof
.
This configuration example applies to OpenLDAP with the memberOf overlay slapo-memberof
and is similar to the Active Directory example.
Provisioned objects, in SPMLv2 terms, consist of identifiers, attributes (probably), and references (maybe) to the identifiers of other objects, which are most likely located on the same provisioning target.
Identifiers consist of a string ID, a target ID, and possibly a container ID. We consider a container ID to be similar to an ldap base dn. A container ID is itself an identifier, recursing potentially indefinitely.
Name value pairs. Probably multi-valued. Case sensitive names and values. We return values in the same order as they were given to us.
A reference refers to the identifier of another object. It consists of two identifiers, the "from object" and the "to object". A node in a directed graph. Directional.
ldap.properties
Configure the default search base DN to match your directory :
edu.vt.middleware.ldap.baseDn = dc=example,dc=edu |
Configure authentication and encryption :
edu.vt.middleware.ldap.serviceUser=cn=Manager,dc=example,dc=edu edu.vt.middleware.ldap.serviceCredential=secret |
Configure the default base DN (container) for people and groups :
# The base DN for groups. edu.internet2.middleware.psp.groupsBaseDn = ou=groups,dc=example,dc=edu # The base DN for people. edu.internet2.middleware.psp.peopleBaseDn = ou=people,dc=example,dc=edu |
ldap.properties
Configure the name of the Grouper stem to be provisioned :
# The base Grouper stem to be provisioned. edu.internet2.middleware.psp.baseStem=edu |
sources.xml
Configure Grouper to look for subjects in your LDAP directory by using the ldap source adapter. Connection information is defined in ldap.properties
.
<source adapterClass="edu.internet2.middleware.subject.provider.LdapSourceAdapter"> <id>ldap</id> <name>LdapSourceAdapter</name> <type>person</type> <init-param> <param-name>ldapProperties_file</param-name> <param-value>ldap.properties</param-value> </init-param> |
Configure the base DN to match your directory in the various search configuration elements :
<param-name>base</param-name> <param-value>ou=people,dc=example,dc=edu</param-value> |
Copy sources.xml
and ldap.properties
from the Grouper API to the Grouper UI.
cp grouper.apiBinary-2.1.0/conf/ldap.properties grouper.ui-2.1.0/dist/grouper/WEB-INF/classes/ cp grouper.apiBinary-2.1.0/conf/sources.xml grouper.ui-2.1.0/dist/grouper/WEB-INF/classes/ |
Copy vt-ldap.jar
from the Grouper API to the Grouper UI :
cp grouper.apiBinary-2.1.0/lib/custom/vt-ldap-3.3.4.jar grouper.ui-2.1.0/dist/grouper/WEB-INF/lib |
The Grouper UI will not load unless you edit |
Comment out or remove the psp specific search result handlers in ldap.properties
in your Grouper UI installation :
# edu.vt.middleware.ldap.searchResultHandlers=edu.internet2.middleware.psp.ldap.QuotedDnResultHandler,... |
In a flat
structure all groups are provisioned under a single base DN (container ID). A flat
group's ldap RDN is its Grouper name or displayName.
In a bushy
structure groups are provisioned hierarchically, with stems as branches (ldap organizationalUnits) in the tree. A bushy
group's RDN is its Grouper extension or displayExtension.
The flat
or bushy
structure is important when calculating ldap DNs and RDNs in the attribute resolver configuration psp-resolver.xml
.
<!-- Group identifier and attributes. --> <!-- The LDAP DN of a group. For example, "cn=groupExtension,ou=stem,ou=groups,dc=edu". --> <resolver:AttributeDefinition id="groupDn" xsi:type="psp-grouper-ldap:LdapDnFromGrouperNamePSOIdentifier" structure="${edu.internet2.middleware.psp.structure}" sourceAttributeID="name" rdnAttributeName="cn" base="${edu.internet2.middleware.psp.groupsBaseDn}"> <resolver:Dependency ref="GroupDataConnector" /> <resolver:Dependency ref="DeleteGroupChangeLogDataConnector" /> <resolver:Dependency ref="UpdateGroupChangeLogDataConnector" /> </resolver:AttributeDefinition> <!-- The value of the group "cn" attribute is the group extension. --> <!-- If the group DN structure is "bushy" the sourceAttributeID should be "extension". --> <!-- If the group DN structure is "flat" the sourceAttributeID should be "name". --> <resolver:AttributeDefinition id="cn" xsi:type="ad:Simple" sourceAttributeID="${edu.internet2.middleware.psp.cnSourceAttributeID}"> <resolver:Dependency ref="GroupDataConnector" /> </resolver:AttributeDefinition> |
In an attempt to make life easier for deployers, the flat
or bushy
structure and corresponding group RDN source attribute ID are configurable via macro replacement in ldap.properties
.
edu.internet2.middleware.psp.structure=flat edu.internet2.middleware.psp.cnSourceAttributeID=name # edu.internet2.middleware.psp.structure=bushy # edu.internet2.middleware.psp.cnSourceAttributeID=extension |
To enable change log provisioning :
changeLog.consumer.psp.class = edu.internet2.middleware.psp.grouper.PspChangeLogConsumer |
To schedule when the change log is processed :
changeLog.consumer.psp.quartzCron = 0 * * * * ? |
To run full synchronizations periodically (by default every day at 5am) :
changeLog.psp.fullSync.class = edu.internet2.middleware.psp.grouper.PspChangeLogConsumer changeLog.psp.fullSync.quartzCron = 0 0 5 * * ? |
To run a full synchronization job at loader startup :
changeLog.psp.fullSync.runAtStartup = true |
The Quartz cron string documentation is here.
Tune Subject API caching in grouper.ehcache.xml
.
Tune maxElementsInMemory
to be greater than or equal to the number of subjects.
Tune timeToIdleSeconds
and timeToLiveSeconds
...
As of Grouper version 2.1.0, which uses Ehcache 2.4, statistics
needs to be "true"
.
<!-- Subject resolving caching --> <!-- @see CachingResolver#find(...) --> <cache name="edu.internet2.middleware.grouper.subj.CachingResolver.Find" maxElementsInMemory="5000" eternal="false" timeToIdleSeconds="30" timeToLiveSeconds="120" overflowToDisk="false" statistics="true" /> <!-- @see CachingResolver#findAll(...) --> <cache name="edu.internet2.middleware.grouper.subj.CachingResolver.FindAll" maxElementsInMemory="5000" eternal="false" timeToIdleSeconds="30" timeToLiveSeconds="120" overflowToDisk="false" statistics="true" /> <!-- @see CachingResolver#findByIdentifier(...) --> <cache name="edu.internet2.middleware.grouper.subj.CachingResolver.FindByIdentifier" maxElementsInMemory="5000" eternal="false" timeToIdleSeconds="30" timeToLiveSeconds="120" overflowToDisk="false" statistics="true" /> <!-- @see CachingResolver#findByIdOrIdentifier(...) --> <cache name="edu.internet2.middleware.grouper.subj.CachingResolver.FindByIdOrIdentifier" maxElementsInMemory="5000" eternal="false" timeToIdleSeconds="30" timeToLiveSeconds="120" overflowToDisk="false" statistics="true" /> |
You may want to change the Grouper log file appenders in grouper.apiBinary-2.1.0/conf/log4j.properties
.
log4j.appender.grouper_error = org.apache.log4j.DailyRollingFileAppender log4j.appender.grouper_error.File = ${grouper.home}logs/grouper_error.log log4j.appender.grouper_error.DatePattern = '.'yyyy-MM-dd log4j.appender.grouper_event = org.apache.log4j.DailyRollingFileAppender log4j.appender.grouper_event.File = ${grouper.home}logs/grouper_event.log log4j.appender.grouper_event.DatePattern = '.'yyyy-MM-dd |
The following changes are necessary to support provisioning Grouper API versions prior to 2.1.0 with the psp. |
1. For versions prior to 2.1.0, there is a bug which will throw a NullPointerException if the following is not present in sources.xml
:
<search> <searchType>searchSubjectByIdentifierAttributes</searchType> <param> <param-name>filter</param-name> <param-value> (&(uid=%TERM%)(objectclass=person)) </param-value> </param> <param> <param-name>scope</param-name> <param-value>SUBTREE_SCOPE</param-value> </param> <param> <param-name>base</param-name> <param-value>ou=people,dc=example,dc=edu</param-value> </param> </search> |
2. For versions prior to 2.1.0, the location of ldap.properties
specified in sources.xml
must be an absolute path. For versions 2.1.0 or later, the location of ldap.properties
may be an absolute path or in your Java classpath.
For example, Grouper API version 2.0.3 requires an absolute path to ldap.properties in sources.xml
:
<init-param> <param-name>ldapProperties_file</param-name> <param-value>/opt/grouper/2.0.3/grouper.apiBinary-2.0.3/conf/ldap.properties</param-value> </init-param> |
For Grouper UI versions prior to 2.1.0, the path to ldap.properties
specified in sources.xml
will be different than in the Grouper API since the psp specific search result handlers must be commented out or removed in the Grouper UI :
<init-param> <param-name>ldapProperties_file</param-name> <param-value>/opt/grouper/2.0.3/grouper.ui-2.0.3/dist/grouper/WEB-INF/classes/ldap.properties</param-value> </init-param> |
3. For Grouper API versions prior to 2.1.0, the ldap source adapter in subject.jar
does not provide the method which allows the psp to re-use the same ldap connection as the subject source. You will need to copy lib/grouper/subject.jar
from the Grouper 2.1.0 API distribution to your pre-2.1.0 Grouper API installation.
The psp configuration files are :
|
Maps source attributes returned by a Shibboleth attribute resolver to target objects. |
|
The Shibboleth attribute resolver configuration file. |
|
Configures Shibboleth attribute resolver services, including provisioning targets. |
|
Bootstraps the Shibboleth attribute resolver. |
The configuration files for Grouper ldap subjects are :
|
Configures Grouper subjects. |
|
The vt-ldap configuration. Also used for macro replacement in the psp configuration files. |
Before you can provision anything from Grouper to ldap or anywhere else, you will need to create the corresponding objects in Grouper using the UI, API, GSH, WS, loader, import, etc.
To calculate how a group should be provisioned :
bin/gsh.sh -psp -calc edu:group |
To diff the current and correct provisioning of a group :
bin/gsh.sh -psp -diff edu:group |
To provision or synchronize a group :
bin/gsh.sh -psp -sync edu:group |
To provision in real-time triggered by the Grouper change log, enable the psp consumer in grouper-loader.properties
and run the loader via
bin/gsh.sh -loader |
Institution |
Subject Source |
Number of Subjects |
Subject ID |
---|---|---|---|
LIGO |
LDAP |
1,000 |
dn: employeeNumber=882,ou=people,dc=ligo,dc=org |
Penn State |
LDAP |
165,000 |
dn:uid=xyx123,dc=psu,dc=edu |
UCLA |
LDAP |
40,000 |
|
UMontreal |
LDAP |
120,000 |
sAMAccountName (value same as cn) |
UVienna |
Undecided |
155,000 |
cn, uid |
UWMadison |
|
|
|
Institution |
Target |
Implementation |
---|---|---|
LIGO |
LDAP |
OpenLDAP 2.4.x |
Penn State |
LDAP |
IBM Tivoli Directory Server |
UCLA |
LDAP |
Sun Java System Directory Server Enterprise Edition 6.3.1 |
UMontreal |
LDAP |
Active Directory |
UVienna |
LDAP |
Active Directory, OpenLDAP |
UWMadison |
|
|
The groups that a member is a member of may be provisioned to the memberOf attribute. Some LDAP implementations, such as Active Directory, automatically maintain the memberOf attribute. OpenLDAP maintains the memberOf attribute automatically via the memberOf overlay. The value of the memberOf attribute is typically a group DN.
Institution |
memberOf for members (people) |
memberOf for groups |
---|---|---|
LIGO |
+ |
|
Penn State |
+ |
|
UCLA |
- |
|
UMontreal |
automatic (Active Directory) |
automatic (Active Directory) |
UVienna |
automatic (Active Directory), OpenLDAP+memberOf |
automatic (Active Directory), OpenLDAP+memberOf |
UWMadison |
+ |
|
The eduMember objectClass defines the isMemberOf and hasMember attributes, whose values are identifiers which are not DNs.
Institution |
isMemberOf |
hasMember |
---|---|---|
LIGO |
+ |
+ |
Penn State |
+ |
+ |
UCLA |
uclaIsMemberOf |
uclaHasMember |
UMontreal |
- |
- |
UVienna |
- |
- |
UWMadison |
+ |
+ |
The eduCourse objectClass defines course related attributes.
Institution |
eduCourse |
---|---|
LIGO |
- |
Penn State |
- |
UCLA |
- |
UMontreal |
- |
UVienna |
- |
UWMadison |
+ |
The group provisioning structure may be either flat
or bushy
. A flat
structure provisions all groups into a single container. A bushy
structure provisions groups hierarchically.
For example, the DN of a group with name 'edu:stem:group' in a flat
structure looks like :
dn: cn=edu:stem:group,ou=groups,dc=example,dc=edu
while the DN of a group with name 'edu:stem:group' in a bushy
structure looks like :
dn: cn=group,ou=stem,ou=edu,ou=groups,dc=example,dc=edu
Institution |
Structure (flat or bushy) |
---|---|
LIGO |
bushy |
Penn State |
flat |
UCLA |
flat |
UMontreal |
bushy |
UVienna |
? |
UWMadison |
flat |
Given groupA with memberA and groupB with memberB :
dn : cn=groupA,ou=groups member: cn=memberA,ou=people dn: cn=groupB,ou=groups member: cn=memberB,ou=people |
If groupB is added as a member to groupA, how do you want groupA to be provisioned :
everything :
dn : cn=groupA,ou=groups member: cn=memberA,ou=people member: cn=memberB,ou=people member: cn=groupB,ou=people |
immediate :
dn : cn=groupA,ou=groups member: cn=memberA,ou=people member: cn=groupB,ou=people |
The everything membership structure handles applications which may not support nested groups and represents the nested structure of the group memberships.
The same membership structure applies to memberOf :
everything :
dn: cn=memberB,ou=people memberOf: cn=groupB,ou=groups memberOf: cn=groupA,ou=groups |
immediate :
dn: cn=memberB,ou=people memberOf: cn=groupB,ou=groups |
Institution |
member |
memberOf |
---|---|---|
LIGO |
everything |
everything |
Penn State |
|
|
UCLA |
|
|
UMontreal |
immediate |
immediate |
UVienna |
everything |
everything |
UWMadison |
|
|