LDAP Plugin

Skip to end of metadata
Go to start of metadata

Plugin Information

Plugin ID ldap Changes In Latest Release
Since Latest Release
Latest Release
Latest Release Date
Required Core
Dependencies
1.11 (archives)
Oct 03, 2014
1.566
mailer (version:1.8)
Source Code
Issue Tracking
Pull Requests
Maintainer(s)
GitHub
Open Issues
Pull Requests
n/a (id: kohsuke)
Usage Installations 2014-Jul 84833
2014-Aug 85097
2014-Sep 89701
2014-Oct 92923
2014-Nov 92971
2014-Dec 93108
2015-Jan 97369
2015-Feb 98324
2015-Mar 106730
2015-Apr 106360
2015-May 106412
2015-Jun 112277

Note: This plugin was part of the Jenkins core until 1.468. After that, it was split out into a separately-updateable plugin. However, for backwards compatibility purposes, subsequent core releases still bundle it. If you do not use this plugin at all, you can simply disable it.

Configuration

Select LDAP for the Security Realm. You will most likely need to configure some of the Advanced options. There is on-line help available for each option. 

Server

Specify the name of the LDAP server host name (like ldap.acme.org).

If your LDAP server uses a port other than 389 (which is the standard for LDAP), you can also append a port number here, like ldap.acme.org:1389.

To connect to LDAP over SSL (AKA LDAPS), specify it with the ldaps:// protocol, like ldaps://ldap.acme.org or ldaps://ldap.acme.org:1636 (if the port is other than the default 636).

As of version 1.6, you can specify a list of servers separated by whitespace to provide a fallback if the first server is unavailable, e.g. ldap1.acme.org ldap2.acme.org:1389 or ldaps://ldap1.acme.org:1636 ldap1.acme.org:1389 ldap://ldap2.acme.org ldap3.acme.org

Root DN

For authenticating user and determing the roles given to this user, Jenkins performs multiple LDAP queries.

Since an LDAP database is conceptually a big tree and the search is performed recursively, in theory if we can start a search starting at a sub-node (as opposed to root), you get a better performance because it narrows down the scope of a search.

This field specifies the DN of such a subtree.

But in practice, LDAP servers maintain an extensive index over the data, so specifying this field is rarely necessary — you should just let Jenkins figure this out by talking to LDAP.

If you do specify this value, the field normally looks something like dc=acme,dc=org

User search base

One of the searches Jenkins does on LDAP is to locate the user record given the user name.

If you specify a relative DN (from the root DN) here, Jenkins will further narrow down searches to the sub-tree.

But in practice, LDAP servers maintain an extensive index over the data, so specifying this field is rarely necessary.

If you do specify this value, the field normally looks something like ou=people

User search filter

One of the searches Jenkins does on LDAP is to locate the user record given the user name.

 This field determines the query to be run to identify the user record.

The query is almost always uid={0} as per defined in RFC 2798, so in most cases you should leave this field empty and let this default kick in.

If your LDAP server doesn't have uid or doesn't use a meaningful uid value, try mail={0}, which lets people login by their e-mail address.

If you do specify a different query, specify an LDAP query string with marker token {0}, which is to be replaced by the user name string entered by the user.

Group search base

One of the searches Jenkins does on LDAP is to locate the list of groups for a user.

This field determines the query to be run to identify the organizational unit that contains groups.

The query is almost always ou=groups so try that first, though this field may be left blank to search from the root DN.

If login attempts result in "Administrative Limit Exceeded" or similar error, try to make this setting as specific as possible for your LDAP structure, to reduce the scope of the query.

If the error persists, you may need to change the Group membership filter from the default of (| (member={0}) (uniqueMember={0}) (memberUid={1})) to a query only of the field used in your LDAP for group membership, such as: (member={0}).

You will need to login and logout in order to verify that your group membership is retained with a modified group membership filter.

Group search filter

When Jenkins is asked to determine if a named group exists, it uses a default filter of:
(& (cn={0}) (| (objectclass=groupOfNames) (objectclass=groupOfUniqueNames) (objectclass=posixGroup)))

relative to the Group search base to determine if there is a group with the specified name ({0} is substituted by the name being searched for.)

If you know your LDAP server only stores group information in one specific object class, then you can improve group search performance by restricting the filter to just the required object class.

Note: if you are using the LDAP security realm to connect to Active Directory (as opposed to using the Active Directory plugin's security realm) then you will need to change this filter to:
(& (cn={0}) (objectclass=group) )

Note: if you leave this empty, the default search filter will be used.

Group membership

When Jenkins resolves a user, the next step in the resolution process is to determine the LDAP groups that the user belongs to.

There is an extension point for providing a strategy to resolve the LDAP groups that the user belongs to. There are two implementations provided in the LDAP plugin:

  • Search for groups containing user (default)
  • Parse user attribute for list of groups

Search for groups containing user

The group membership filter field controls the search filter that is used to determine group membership.

If left blank, the default filter will be used. The default default filter is: (| (member={0}) (uniqueMember={0}) (memberUid={1})). Irrespective of what the default is, setting this filter to a non-blank value will determine the filter used.

You are normally safe leaving this field unchanged, however for large LDAP servers where you are seeing messages such as "OperationNotSupportedException - Function Not Implemented", "Administrative Limit Exceeded" or similar periodically when trying to login, then that would indicate that you should change to a more optimum filter for your LDAP server, namely one that queries only the required field, such as: (member={0})

The LDAP server may be able to use query hints to optimize the search. For example:

  • If all the groups you are interested in are within a specific subtree, adding the subtree information to the filter should improve performance.
  • Active Directory's query optimizer can make significant optimizations if it knows that the object category is group: (&(objectCategory=group)(member={0})) this may be relevant if using Active Directory's matching rule in chain extension, e.g. (&(objectCategory=group)(member:1.2.840.113556.1.4.1941:={0}))

Note: in this field there are two available substitutions:
{0} - the fully qualified DN of the user
{1} - the username portion of the user

Parse user attribute for list of groups


Some LDAP servers can provide a memberOf attribute within the User's record:

  • Active Directory
  • OpenLDAP with the memberof overlay active (untested, and as memberof is an operational attribute in OpenLDAP it must be explicitly requested, so likely some hacking of LDAPSecurityRealm.groovy required)
  • (If you know of others please provide details here)

This attribute can be used to simplify the group search and return the group membership immediately without a second LDAP query. Note, however, that this may result in only direct group membership being supported.

The group membership attribute field controls the attribute name that is used to determine the groups to which a user belongs.

Manager DN and Manager Password

If your LDAP server doesn't support anonymous binding (IOW, if your LDAP server doesn't even allow a query without authentication), then Jenkins would have to first authenticate itself against the LDAP server, and Jenkins does that by sending "manager" DN and password.

A DN typically looks like CN=MyUser,CN=Users,DC=mydomain,DC=com although the exact sequence of tokens depends on the LDAP server configuration.

It can be any valid DN as long as LDAP allows this user to query data.

This configuration is also useful when you are connecting to Active Directory from a Unix machine, as AD doesn't allow anonymous bind by default. But if you can't figure this out, you can also change AD setting to allow anonymous bind. 

Disable LDAP Email resolver

Controls whether LDAP will be used to try and resolve the email addresses of users.

Enable cache

Some LDAP servers may be slow, or rate limit client requests.

In such cases enabling caching may improve performance of Jenkins with the risk of delayed propagation of user changes from LDAP and increased memory usage on the master.

Note: The default configuration is to leave the cache turned off.

Environment Properties

As of 1.7 of the LDAP plugin, you can now specify additional Environment Properties to provide the backing Java LDAP client API. See Oracle's documentation for details of what properties are available and what functionality they provide. As a minimum you should strongly consider providing the following

Property Name Description
com.sun.jndi.ldap.connect.timeou
This is the socket connection timeout in milliseconds. If your LDAP servers are all close to your Jenkins server you can probably set a small value, e.g. 5000 milliseconds. Setting a value smaller that this may result in excessive timeouts due to the TCP/IP connection establishment retry mechanism.
com.sun.jndi.ldap.read.timeout This is the socket read timeout in milliseconds. If your LDAP queries are all fast you can probably set a low value. The value is ignored if the Jenkins Master is running on Java 1.5. A reasonable default is 60000 milliseconds.

Troubleshooting

The following Groovy script can be useful when trying to determine whether you have group search configured correctly:

    String[] names = ["a group name","a user name","a name that does not exist"];
    for (name in names) {
      println("Checking the name '" + name + "'...")
      try {
        println("  It is a USER: " + Jenkins.instance.securityRealm.loadUserByUsername(name))
      } catch (Exception e) {
          try {
            println("  It is a GROUP: " + Jenkins.instance.securityRealm.loadGroupByGroupname(name))
            println("")
            continue
          } catch (Exception e1) {
            println("  It is NOT a group, reason: " + e1.getMessage())
          }
        println("  It is NOT a user, reason: " + e.getMessage())
      }
      println("");
    }
  • If login attempts result in "OperationNotSupportedException - Function Not Implemented", "Administrative Limit Exceeded" or similar error, the LDAP query to determine the group membership for the user may be triggering this. First try setting the "Group search base" setting as specific as possible for your LDAP structure, to reduce the scope of the query. If the error persists, you may need to edit the WEB-INF/security/LDAPBindSecurityRealm.groovy file that is included in jenkins.war. Change the line with groupSearchFilter = "(| (member={0}) (uniqueMember={0}) (memberUid={1}))"; to query only of the field used in your LDAP for group membership, such as groupSearchFilter = "(member={0})"; (then restart Jenkins).
  • The LDAP groups were available in Jenkins in the format of ROLE_Uppercasedgroupname, so the developers ldap group would be ROLE_Developers in Jenkins, but since 1.404 they are available as is: no prefix or upper casing,
  • Since Jenkins 1.468, this has been moved to a plugin. The LDAPBindSecurityRealm.groovy file is therefore part of the ldap.jpi file. You can find the default template at $JENKINS_HOME/plugins/ldap/WEB-INF/classes/hudson/security/LDAPBindSecurityRealm.groovy. That file will be recreated from the ldap.jpi file every time Jenkins starts, so if you need to override the defaults, the correct way is to just copy the template file to $JENKINS_HOME/LDAPBindSecurityRealm.groovy. The $JENKINS_HOME/LDAPBindSecurityRealm.groovy file is re-read every time the security components are reconfigured, so it should just be a case of re-saving the security configuration to force the file to be re-read.
  • If you are using this plugin and not the Active Directory plugin to connect to Active Directory, you will need to change the Group Search Filter to filter to: (& (cn={0}) (objectclass=group) ) and change the Group Membership Filter to: (member={0}). If you want AD to return nested group membership then change the Group Membership Filter to: (member:1.2.840.113556.1.4.1941:={0})

Performance Tuning

Here is a checklist to help improve performance:

  • Ensure you are using the very latest version of the LDAP plugin
  • Ensure you have enabled caching. Start with the cache size to just greater than your anticipated maximum concurrent users and set the TTL to the longest time interval you are comfortable with... (i.e. how long before a password change gets picked up... in most cases 5 or 10 minutes is a good TTL)

Those two changes should give you an immediate significant performance boost (even with a TTL of 30s as long as the cache size is larger than max anticipated concurrent users... but a longer TTL is better)

  • Next up is to ensure that you have got the correct most specific user search base and group search base defined for your LDAP tree. Getting this right has two side-effects... you get faster results to your queries; and your LDAP server admin people will thank you for reducing the load on their server by a significant amount.
  • Finally, you should ensure that you have defined specific queries for the user search filter and group search filter... the user one is usually fine as is... the group one is, by default, a combination of typical queries. A significant performance improvement can be achieved by switching from the default or filter of (& (cn={0}) (| (objectclass=groupOfNames) (objectclass=groupOfUniqueNames) (objectclass=posixGroup))) to the correct for your LDAP tree query, i.e. it would be one of (& (cn={0}) (objectclass=groupOfNames)),(& (cn={0}) (objectclass=groupOfUniqueNames)) or (& (cn={0}) (objectclass=posixGroup)). (...and if it is not one of them then your LDAP server is most likely Active Directory and Kohsuke makes me ask why you are using the LDAP plugin and not the Active Directory plugin in that case! Note that JENKINS-16429 might be a good reason to favour the LDAP plugin over the Active Directory plugin, but if that issue is resolved by the time you are reading this then there should be no reason to pick the LDAP plugin over the Active Directory plugin)

Tips and Tricks

If you are using the LDAP plugin to connect to Active Directory you should probably read this page of AD syntax notes. Pay special attention to Notes 10 and 19. The following settings are reported to work with Active Directory and nested groups, though they should carry a warning that they may impact login performance and they have not been tested for completeness:

  • User search filter: sAMAccountName={0}
  • Group search filter: (&(objectclass=group)(cn={0}))
  • Group membership, one of
    • Search for groups containing user (if nested group membership required)
      • Group membership filter: (&(objectCategory=group)(member:1.2.840.113556.1.4.1941:={0}))
    • Parse user attribute for list of groups (if nested group membership not required this will be faster)
      •  Group membership attribute: memberOf

Version History

Version 1.11 (3rd Oct 2014)

  • Performance improvements especially in the presence of lots of requests with HTTP basic auth.

Version 1.10.2 (23rd May 2014)

  • Fixed another NPE in FromUserRecordLDAPGroupMembershipStrategy

Version 1.10.1 (23rd May 2014)

  • Fixed NPE in FromUserRecordLDAPGroupMembershipStrategy.

Version 1.10 (22nd May 2014)

  • Turned the group membership lookup into a strategy. There are now two strategies, the default "look up groups containing the user" strategy and an experimental new strategy which looks for an attribute in the user's LDAP record that contains a list of DNs of the groups that the user belongs to. Rumour has it that this second strategy may actually provide faster performance for Active Directory, but as the person who wrote this code does not have an Active Directory instance to test against - until some kind soul tests, confirms and edits this text to remove the assertion that this is a rumour - using the new strategy is Caveat emptor.

[Update 23/05/2014] Some kind testers have confirmed that the new strategy seems to work against Active Directory... but as those testers did not have performance issues to start with, again it is just a rumour that there is a performance increase! Version 1.10.2 is recommended to fix two non-critical but annoying NPEs with the new strategy

Version 1.9 (9th May 2014)

  • Added some interim hacks to work around JENKINS-22247. Setting the temporary system properties 
    hudson.security.LDAPSecurityRealm.forceUsernameLowercase=true
    

    and 

    hudson.security.LDAPSecurityRealm.forceGroupnameLowercase=true
    

    will enable these hacks. These system properties will be removed in a future version once the core issue has been resolved.

  • Modernised the configuration screen Jelly to use current form-binding.
  • The manager password is now correctly encrypted using Secret. This is a downgrade breaking change. WARNING! If you upgrade to 1.9 and then downgrade, the manager password may be lost from your configuration. 

Version 1.8 (17th Jan 2014)

Older versions

Look at the commit history for details of the changes in older versions. (SC - If you want to update this wiki page with those details as a service to the community please feel free to do so, this page was missing any version history, so I added at least some history)

Labels

Edit
plugin-user plugin-user Delete
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.

Add Comment