Chapter 9 LDAP Basic Components and Concepts (Overview)

This chapter covers the following topics.

• LDAP Data Interchange Format (LDIF)

• Using Fully Qualified Domain Names With LDAP

• Default Directory Information Tree (DIT)

• Default LDAP Schema

• Service Search Descriptors (SSDs) and Schema Mapping

• LDAP Client Profiles

• ldap_cachemgr Daemon

• LDAP Naming Services Security Model

LDAP Data Interchange Format (LDIF)

LDIF is a text-based format for describing directory service entities and their attributes. Using LDIF format you can move information from one directory to another with commands such as ldapadd and ldapmodify. The following are examples of LDIF format for each service. Use ldaplist(1) with the-l option to display the following information.

% ldaplist -l hosts myhost

hosts

dn: cn=myhost+ipHostNumber=7.7.7.115,ou=Hosts,dc=mydc,dc=mycom,dc=com
cn: myhost
iphostnumber: 7.7.7.115
objectclass: top
objectclass: device
objectclass: ipHost
description: host 1 - floor 1 - Lab a - building b

% ldaplist -l passwd user1

passwd

dn: uid=user1,ou=People,dc=mydc,dc=mycom,dc=com
uid: user1
cn: user1
userpassword: {crypt}duTx91g7PoNzE
uidnumber: 199995
gidnumber: 20
gecos: Joe Smith [New York]
homedirectory: /home/user1
loginshell: /bin/csh
objectclass: top
objectclass: shadowAccount
objectclass: account
objectclass: posixAccount

% ldaplist -l services name

services

dn: cn=name+ipServiceProtocol=udp,ou=Services,dc=mydc,dc=mycom,dc=com
cn: name
cn: nameserver
ipserviceprotocol: udp
ipserviceport: 42
objectclass: top
objectclass: ipService

% ldaplist -l group mygroup

group

dn: cn=mygroup,ou=Group,dc=mydc,dc=mycom,dc=com
cn: mygroup
gidnumber: 4441
memberuid: user1
memberuid: user2
memberuid: user3
userpassword: {crypt}duTx91g7PoNzE
objectclass: top
objectclass: posixGroup

% ldaplist -lnetgroup mynetgroup

netgroup

cn=mynetgroup,ou=netgroup,dc=central,dc=sun,dc=com objectclass=nisNetgroup
-objectclass: -top
-cn: -mynetgroup
-nisnetgrouptriple: -(user1..mydc.mycom.com,-,) nisnetgrouptriple=(user1.,-,)
-membernisnetgroup: -mylab

% ldaplist -l networks 200.20.20.0

networks

dn: ipNetworkNumber=200.20.20.0,ou=Networks,dc=mydc,dc=mycom,dc=com
cn: mynet-200-20-20
ipnetworknumber: 200.20.20.0
objectclass: top
objectclass: ipNetwork
description: my Lab Network
ipnetmasknumber: 255.255.255.0

% ldaplist -l netmasks 201.20.20.0

netmasks

dn: ipNetworkNumber=201.20.20.0,ou=Networks,dc=mydc,dc=mycom,dc=com
cn: net-201
ipnetworknumber: 201.20.20.0
objectclass: top
objectclass: ipNetwork
description: my net 201
ipnetmasknumber: 255.255.255.0

% ldaplist -l rpc ypserv

rpc

dn: cn=ypserv,ou=Rpc,dc=mydc,dc=mycom,dc=com
cn: ypserv
cn: ypprog
oncrpcnumber: 100004
objectclass: top
objectclass: oncRpc

% ldaplist -l protocols tcp

protocols

dn: cn=tcp,ou=Protocols,dc=mydc,dc=mycom,dc=com
cn: tcp
ipprotocolnumber: 6
description: transmission control protocol
objectclass: top
objectclass: ipProtocol

% ldaplist -l bootparams myhost

bootparams

dn: cn=myhost,ou=Ethers,dc=mydc,dc=mycom,dc=com
bootparameter: root=boothost:/export/a/b/c/d/e
objectclass: top
objectclass: device
objectclass: bootableDevice
cn: myhost

% ldaplist -l ethers myhost

ethers

dn: cn=myhost,ou=Ethers,dc=mydc,dc=mycom,dc=com
macaddress: 8:1:21:71:31:c1
objectclass: top
objectclass: device
objectclass: ieee802Device
cn: myhost

% ldaplist -l publickey myhost

publickey

dn: cn=myhost+ipHostNumber=200.20.20.99,ou=Hosts,dc=mydc,dc=mycom,dc=com
cn: myhost
iphostnumber: 200.20.20.99
description: Joe Smith
nispublickey: 9cc01614d929848849add28d090acdaa1c78270aeec969c9
nissecretkey: 9999999998769c999c39e7a6ed4e7afd687d4b99908b4de99
objectclass: top
objectclass: NisKeyObject
objectclass: device
objectclass: ipHost

% ldaplist -l aliases myname

aliases

dn: mail=myname,ou=aliases,dc=mydc,dc=mycom,dc=com
cn: myname
mail: myname
objectclass: top
objectclass: mailgroup
mgrprfc822mailmember: my.name

Using Fully Qualified Domain Names With LDAP

Unlike NIS or NIS+ clients, an LDAP client always returns a fully qualified domain name (FQDN) for a host name. The LDAP FQDN is similar to the FQDN returned by DNS. For example, suppose your domain name is the following:

west.example.net

Both gethostbyname() and getnameinfo() return the FQDN version when looking up the host name server:

server.west.example.net

Also, if you use interface-specific aliases such as server-#, a long list of fully qualified host names are returned. If you are using host names to share file systems or have other such checks, you must account for the checks. For example, if you assume non-FQDNs for local hosts and FQDNs only for remote DNS-resolved hosts, you must account for the difference. If you set up LDAP with a different domain name from DNS, the same host might end up with two different FQDNs, depending on the lookup source.

Default Directory Information Tree (DIT)

By default, Solaris LDAP clients access the information assuming that the DIT has a given structure. For each domain supported by the LDAP server, there is a subtree with an assumed structure. This default structure, however, can be overridden by specifying Service Search Descriptors (SSDs). For a given domain, the default DIT will have a base container that holds a number of well known containers that hold entries for a specific information type. See the following table for the names of these subtrees. (This information can be found in RFC 2307 and others.)

Default LDAP Schema

Schemas are definitions describing what types of information can be stored as entries in an LDAP directory. To support LDAP naming clients, the directory server's schema might need to be extended. Detailed information about IETF and Solaris specific schemas is included in Chapter 14, LDAP General Reference (Reference). The various RFCs can also be accessed on the IETF Web site http://www.ietf.org.

Service Search Descriptors (SSDs) and Schema Mapping

If you use schema mapping, you must do so in a very careful and consistent manner. Make sure the syntax of the mapped attribute is consistent with the attribute it is mapped to. In other words, make sure that single-valued attributes map to single-valued attributes, that the attribute syntaxes are in agreement, and that mapped object classes have the correct mandatory (possibly mapped) attributes.

As previously discussed, LDAP naming services expect, by default, the DIT to be structured in a certain way. If you want, you can instruct the Solaris LDAP naming service to search in other locations than the default locations in the DIT. Additionally, you can specify that different attributes and object classes be used in place of those specified by the default schema. For a list of default filters, see Default Filters Used by LDAP Naming Services.

Description of SSDs

The serviceSearchDescriptor attribute defines how and where an LDAP naming service client should search for information for a particular service. The serviceSearchDescriptor contains a service name, followed by one or more semicolon-separated base-scope-filter triples. These base-scope-filter triples are used to define searches only for the specific service and are searched in order. If multiple base-scope-filters are specified for a given service, then when that service looks for a particular entry, it will search in each base with the specified scope and filter.

The default location is not searched for a service (database) with an SSD unless it is included in the SSD. Unpredictable behavior will result if multiple SSDs are given for a service.

In the following example, the Solaris LDAP naming service client performs a one-level search in ou=west,dc=example,dc=com followed by a one-level search in ou=east,dc=example,dc=com for the passwd service. To look up the passwd data for a user's username, the default LDAP filter (&(objectClass=posixAccount)(uid=username)) is used for each BaseDN.

serviceSearchDescriptor: passwd:ou=west,dc=example,dc=com;ou=east,
dc=example,dc=com 

In the following example, the Solaris LDAP naming service client would perform a subtree search in ou=west,dc=example,dc=com for the passwd service. To look up the passwd data for user username, the subtree ou=west,dc=example,dc=com would be searched with the LDAP filter (&(fulltimeEmployee=TRUE)(uid=username)).

serviceSearchDescriptor: passwd:ou=west,dc=example,
dc=com?sub?fulltimeEmployee=TRUE

It is also possible to associate multiple containers with a particular service type. In the following example, the service search descriptor specifies searching for the password entries in three containers.

Note that a trailing ',' in the example implies that the defaultSearchBase is appended to the relative base in the SSD.

defaultSearchBase: dc=example,dc=com
serviceSearchDescriptor: \
passwd:ou=myuser,;ou=newuser,;ou=extuser,dc=example,dc=com

Attribute Map

The Solaris LDAP naming service allows one or more attribute names to be remapped for any of its services. (The Solaris LDAP client uses the well-known attributes documented in Chapter 14, LDAP General Reference (Reference).) If you map an attribute, you must be sure that the attribute has the same meaning and syntax as the original attribute. Note that mapping the userPassword attribute might cause problems.

There are a couple of reasons you might want to use schema mappings.

• You want to map attributes in an existing directory server

• If you have user names that differ only in case, you must map the uid attribute, which ignores case, to an attribute that does not ignore case

The format for this attribute is service:attribute-name=mapped-attribute-name.

If you want to map more than one attribute for a given service, you can define multiple attributeMap attributes.

In the following example, the employeeName and home attributes would be used whenever the uid and homeDirectory attributes would be used for the passwd service.

attributeMap: passwd:uid=employeeName
attributeMap: passwd:homeDirectory=home

There exists one special case where you can map the passwd service's gecos attribute to several attributes. The following is an example.

attributemap: gecos=cn sn title

This maps the gecos values to a space separated list of the cn, sn, and title attribute values.

objectClass Map

The Solaris LDAP naming service allows object classes to be remapped for any of its services. If you want to map more than one object class for a given service, you can define multiple objectclassMap attributes. In the following example, the myUnixAccount object class is used whenever the posixAccount object class is used.

objectclassMap: passwd:posixAccount=myUnixAccount

LDAP Client Profiles

To simplify Solaris client setup, and avoid having to reenter the same information for each and every client, create a single client profile on the directory server. This way, a single profile defines the configuration for all clients configured to use it. Any subsequent change to the profile attributes is propagated to the clients at a rate defined by the refresh interval.

These client profiles should be stored in a well-known location on the LDAP server. The root DN for the given domain must have an object class of nisDomainObject and a nisDomain attribute containing the client's domain. All profiles are located in the ou=profile container relative to this container. These profiles should be readable anonymously.

Client Profile Attributes

The following table shows the Solaris LDAP client's profile attributes, which can be set automatically when you run idsconfig. See Initializing a Client Manually and the idsconfig(1M) man page for information on how to set a client profile manually.

Local Client Attributes

The following table lists the client attributes that can be set locally using ldapclient. See the ldapclient(1M) man page for more information.

Starting in the Solaris 10 10/09 release, the enableShadowUpdate switch is available. For more information, see enableShadowUpdate Switch.

If the BaseDN in an SSD contains a trailing comma, it is treated as a relative value of the defaultSearchBase. The values of the defaultSearchBase are appended to the BaseDN before a search is performed.

ldap_cachemgr Daemon

ldap_cachemgr is a daemon that runs on LDAP client machines. When you start the LDAP client, the ldap_cachemgr daemon is invoked. The daemon performs the following key functions.

• Gains access to the configuration files, running as root

• Refreshes the client configuration information stored in the profiles on the server and pulls this data from the clients

• Maintains a sorted list of active LDAP servers to use

• Improves lookup efficiency by caching some common lookup requests submitted by various clients

• Improves the efficiency of host lookups

• Starting in the Solaris 10 10/09 release, if the enableShadowUpdate switch is set to true, gains access to the configured administrator credential and performs updates to the shadow data

ldap_cachemgr must be running at all times for LDAP naming services to work.

Refer to the ldap_cachemgr(1M) man page for detailed information.

LDAP Naming Services Security Model

Introduction

Solaris LDAP naming services can use the LDAP repository in two different ways. One is as a source of both a naming service and an authentication service. The other is strictly as the source of naming data. This section discusses the concepts of client identity, authentication methods, pam_ldap and pam_unix modules, and account management when the LDAP repository is used as both a naming service and authentication service. This section also discusses the use of LDAP naming services in conjunction with the Kerberos environment (Part VI, Kerberos Service, in System Administration Guide: Security Services) and pam_krb5(5) modules.

Previously, if you enabled pam_ldap account management, all users needed to provide a login password for authentication any time they logged in to the system. Therefore, nonpassword-based logins using tools such as rsh, rlogin, or ssh would fail.

Now, however, pam_ldap(5), when used with Sun Java System Directory Servers DS5.2p4 and newer releases, enables users to log in with rsh, rlogin, rcp and ssh without giving a password.

pam_ldap(5) is now modified to perform account management and retrieve the account status of users without authenticating to Directory Server as the user logging in. The new control to this on Directory Server is 1.3.6.1.4.1.42.2.27.9.5.8, which is enabled by default.

To modify this control for other than default, add Access Control Instructions (ACI) on Directory Server:

dn: oid=1.3.6.1.4.1.42.2.27.9.5.8,cn=features,cn=config
objectClass: top
objectClass: directoryServerFeature
oid:1.3.6.1.4.1.42.2.27.9.5.8
cn:Password Policy Account Usable Request Control
aci: (targetattr != "aci")(version 3.0; acl "Account Usable"; 
     allow (read, search, compare, proxy)
     (groupdn = "ldap:///cn=Administrators,cn=config");)
creatorsName: cn=server,cn=plugins,cn=config
modifiersName: cn=server,cn=plugins,cn=config

If you use Kerberos as your authentication system and integrate it with the LDAP naming system, you will be able to support a single sign on (SSO) environment in your enterprise through Kerberos. You will also be able to use that same identity system when querying LDAP naming data on a per-user or per-host basis.

To access the information in the LDAP repository, clients can first establish identity with the directory server. This identity can be either anonymous or as an object recognized by the LDAP server. Based on the client's identity and the server's access control information (ACI), the LDAP server will allow the client to read or write directory information. For more information on ACIs, consult the Administration Guide for the version of Sun Java System Directory Server that you are using.

If the client is connecting as anything other than anonymous for any given request, the client must prove its identity to the server using an authentication method supported by both the client and the server. Once the client has established its identity, it can then make the various LDAP requests.

When you use pam_ldap there is a distinction between how the naming service and the authentication service (pam_ldap) access the directory. The naming service reads various entries and their attributes from the directory based on predefined identity. The authentication service establishes whether the user has entered the correct password by using that user's name and password to authenticate to the LDAP server. See the pam_ldap(5) man page for more information about the authentication service.

When Kerberos is used to perform authentication, and when authentication in LDAP naming services is also enabled (as is required for per-user mode), Kerberos can provide dual functions. Kerberos authenticates to the server and the Kerberos identity for the principal (user or host) is used to authenticate to the directory. In this way, the same user identity that is used to authenticate to the system is also used to authenticate to the directory for lookups and updates. Administrators can use access control information (ACI) in the directory to limit the results out of the naming service if desired.

Transport Layer Security (TLS)

In order to use TLS for Solaris LDAP naming services, the directory server must use the default ports, 389 and 636, for LDAP and SSL, respectively. If your directory server does not use these ports, you cannot use TLS at this time.

TLS can be used to secure communication between an LDAP client and the directory server, providing both privacy and data integrity. The TLS protocol is a superset of the Secure Sockets Layer (SSL) protocol. Solaris LDAP naming services support TLS connections. Be aware that using SSL adds load to the directory server and the client.

You will need to set up your directory server for SSL. For more information about setting up Sun Java System Directory Server for SSL, see the Administration Guide for the version of Sun Java System Directory Server that you are using. You will also need to set up your LDAP client for SSL.

If using TLS, the necessary security databases must be installed. In particular, the certificate and key database files are needed. For example, if you adopt an older database format from Netscape Communicator, two files, cert7.db and key3.db, are required. Or if you use a new database format from Mozilla, three files, cert8.db, key3.db, and secmod.db are needed. The cert7.db or cert8.dbfile contains trusted certificates. The key3.dbfile contains the client's keys. Even if the LDAP naming service client does not use client keys, this file must be present. The secmod.db file contains the security modules such as the PKCS#11 module. This file is not required if the older format is used.

See Setting Up TLS Security for more information.

Assigning Client Credential Levels

LDAP naming services clients authenticate to the LDAP server according to a client's credential level. LDAP clients can be assigned four possible credential levels with which to authenticate to a directory server.

• anonymous

• proxy

• proxy anonymous

• self (called per user in this document)

Anonymous

If you use anonymous access, you can access only the data that is available to everyone. In anonymous mode, an LDAP BIND operation does not take place. Also, you should consider the security implications. Allowing anonymous access for certain parts of the directory implies that anyone with access to the directory has read access. If you use an anonymous credential level, you need to allow read access to all the LDAP naming entries and attributes.

Allowing anonymous write to a directory should never be done, as anyone could change information in the DIT to which they have write access, including another user's password, or their own identity.

Sun Java System Directory Server allows you to restrict access based on IP addresses, DNS name, authentication method, and time-of-day. You might want to limit access with further restrictions. For more information, see “Managing Access Control” in the Administration Guide for the version of Sun Java System Directory Server that you are using.

Proxy

The client authenticates or binds to the directory using a single proxy account. This proxy account can be any entry that is allowed to bind to the directory. This proxy account needs sufficient access to perform the naming service functions on the LDAP server. The proxy account is a shared-per-system resource. That is, each user logged in to a system using proxy access, including the root user, sees the same results as all other users on that system. You need to configure the proxyDN and proxyPassword on every client using the proxy credential level. The encrypted proxyPassword is stored locally on the client. You can set up different proxies for different groups of clients. For example, you can configure a proxy for all the sales clients to access both the company-wide-accessible and sales directories, while preventing sales clients from accessing human resource directories with payroll information. Or, in the most extreme cases, you can either assign different proxies to each client or assign just one proxy to all clients. A typical LDAP deployment would probably lie between the two extremes. Consider the choices carefully. Too few proxy agents might limit your ability to control user access to resources. However, having too many proxies complicates the setup and maintenance of the system. You need to grant the appropriate rights to the proxy user, depending on your environment. See Credential Storage for information on how to determine which authentication method makes the most sense for your configuration.

If the password changes for a proxy user, you need to update it on every client that uses that proxy user. If you use password aging on LDAP accounts, be sure to turn it off for proxy users.

Be aware that the proxy credential level applies to all users and processes on any given system. If two users need to use different naming policies, they must use different machines, or they must use the per-user authentication model.

In addition, if clients are using a proxy credential to authenticate, the proxyDN must have the same proxyPassword on all of the servers.

Proxy anonymous

Proxy anonymous is a multi-valued entry, in that more than one credential level is defined. A client assigned the proxy anonymous level will first attempt to authenticate with its proxy identity. If the client is unable to authenticate as the proxy user for whatever reason (user lockout, password expired, for example), then the client will use anonymous access. This might lead to a different level of service, depending on how the directory is configured.

Per User

Per-user (self) authentication uses the Kerberos identity (principal) to perform a lookup for each user or each system when authenticating to the directory server. With per-user authentication, the system administrator can use access control instructions (ACI's), access control lists (ACL's), roles, groups or other directory access control mechanisms to grant or deny access to specific naming service data for specific users or systems.

When configuring per-user mode, the configuration value to enable this mode is “self,” which denotes per-user mode.

To use the per-user authentication model, the Kerberos single sign-on service must be deployed. In addition, the one or more directory servers used in the deployment must support SASL and the SASL/GSSAPI authentication mechanism. Because Kerberos expects to use files and DNS for host name lookups, instead of LDAP, DNS should be deployed in this environment. Also, to use per-user authentication, nscd must be enabled. The nscd daemon is not an optional component in this configuration.

enableShadowUpdate Switch

Starting in the Solaris 10 10/09 release, if the enableShadowUpdate switch is set to true on the client, the admin credentials will be used to update the shadow data. Shadow data is stored in the shadowAccount object class on the directory server. Admin credentials are defined by the values of the adminDN and adminPassword attributes, as described in Local Client Attributes. These admin credentials are not used for any other purpose.

Admin credentials have properties similar to Proxy credentials. The exception is that for admin credentials, the user must have all privileges for the zone or have an effective UID of root to read or update the shadow data. Admin credentials can be assigned to any entry that is allowed to bind to the directory. However, do not use the same directory manager identity (cn=Directory Manager) of the LDAP server.

This entry with admin credentials must have sufficient access to read and write the shadow data in the directory. Because the entry is a shared-per-system resource, the adminDN and adminPassword attributes must be configured on every client. The encrypted adminPassword is stored locally on the client. The password uses the same authentication methods that are configured for the client. The admin credentials are used by all users and processes on a given system to read and update the shadow data.

Credential Storage

If you configure a client to use a proxy identity, the client saves its proxyDN and proxyPassword in /var/ldap/ldap_client_cred. For the sake of increased security, this file is restricted to root access only, and the value of proxyPassword is encrypted. While past LDAP implementations have stored proxy credentials in a client's profile, Solaris 9 LDAP naming services do not. Any proxy credentials set using ldapclient during initialization are stored locally. This results in improved security surrounding a proxy's DN and password information. See Chapter 12, Setting Up LDAP Clients (Tasks) for more information on setting up client profiles.

Similarly, if you configure a client to enable shadow data updates, and the client credential level is not self, the client saves its adminDN and adminPassword attributes locally in the /var/ldap/ldap_client_cred file. The value of adminPassword is also encrypted and is used only by the ldap_cachemgr daemon process.

If you configure a client to use per-user authentication, the Kerberos identity and Kerberos ticket information for each principal (each user or host) are used during authentication. In this environment the directory server maps the Kerberos principal to a DN and the Kerberos credentials are used to authenticate to that DN. The directory server can then use its access control instruction (ACI) mechanisms to allow or deny access to naming service data as necessary. In this situation, Kerberos ticket information is used to authenticate to the directory server and the system does not store authentication DNs or passwords on the system. Therefore, for this type of configuration, you do not need to specify the adminDN and adminPassword attributes when the client is initialized with the ldapclient command.

Choosing Authentication Methods

When you assign the proxy or proxy-anonymous credential level to a client, you also need to select a method by which the proxy authenticates to the directory server. By default, the authentication method is none, which implies anonymous access. The authentication method may also have a transport security option associated with it.

The authentication method, like the credential level, may be multivalued. For example, in the client profile you could specify that the client first tries to bind using the simple method secured by TLS. If unsuccessful, the client would try to bind with the sasl/digest-MD5 method. The authenticationMethod would then be tls:simple;sasl/digest-MD5.

LDAP naming services support some Simple Authentication and Security Layer (SASL) mechanisms. These mechanisms allow for a secure password exchange without requiring TLS. However, these mechanisms do not provide data integrity or privacy. See RFC 2222 for information on SASL.

The following authentication mechanisms are supported.

• none

  The client does not authenticate to the directory. This is equivalent to the anonymous credential level.

• simple

  If the client system uses the simple authentication method, it binds to the server by sending the user's password in the clear. The password is thus subject to snooping unless the session is protected by IPsec. The primary advantages of using the simple authentication method are that all directory servers support it and that it is easy to set up.

• sasl/digest-MD5

  The client's password is protected during authentication, but the session is not encrypted. Some directory servers, including Sun Java System Directory Server, also support the sasl/digest-MD5 authentication method. The primary advantage of digest-MD5 is that the password does not go over the wire in the clear during authentication and therefore is more secure than the simple authentication method. See RFC 2831 for information on digest-MD5. digest-MD5 is considered an improvement over cram-MD5 for its improved security.

  When using sasl/digest-MD5, the authentication is secure, but the session is not protected.

  ---

  Note –

  If you are using Sun Java System Directory Server, the password must be stored in the clear in the directory.

  ---

• sasl/cram-MD5

  In this case, the LDAP session is not encrypted, but the client's password is protected during authentication, as authentication is performed by using sasl/cram-MD5.

  See RFC 2195 for information on the cram-MD5 authentication method. cram-MD5 is only supported by some directory servers. For instance, Sun Java System Directory Server does not support cram-MD5.

• sasl/GSSAPI

  This authentication method is used in conjunction with the self credential mode to enable per-user lookups. A per-user nscd assigned to use the client's credentials binds to the directory server using the sasl/GSSAPI method and the client's Kerberos credentials. Access can be controlled in the directory server on a per-user basis.

• tls:simple

  The client binds using the simple method and the session is encrypted. The password is protected.

• tls:sasl/cram-MD5

  The LDAP session is encrypted and the client authenticates to the directory server using sasl/cram-MD5.

• tls:sasl/digest-MD5

  The LDAP session is encrypted and the client authenticates to the directory server using sasl/digest-MD5.

Sun Java System Directory Server requires passwords to be stored in the clear in order to use digest-MD5. If the authentication method is set to sasl/digest-MD5 or tls:sasl/digest-MD5, then the passwords for the proxy user will need to be stored in the clear. Be especially careful that the userPassword attribute has the proper ACIs if it is stored in the clear, so that it is not readable.

The following table summarizes the various authentication methods and their respective characteristics.

Authentication and Services

The authentication method can be specified for a given service in the serviceAuthenticationMethod attribute. The following services currently support this.

• passwd-cmd

  This service is used by passwd(1) to change the login password and password attributes.

• keyserv

  This service is used by the chkey(1) and newkey(1M) utilities to create and change a user's Diffie-Hellman key pair.

• pam_ldap

  This service is used for authenticating users with pam_ldap(5).

  pam_ldap supports account management.

If the service does not have a serviceAuthenticationMethod set, it will default to the value of the authenticationMethod attribute.

In per-user mode, pam_krb5 Service Module (pam Kerberos) is used as the authentication service. ServiceAuthenticationMethod is not needed in this mode of operation.

If the enableShadowUpdate switch is set to true, the ldap_cachemgr daemon binds to the LDAP server by using the authentication method that is defined in the serviceAuthenticationMethod parameter of passwd-cmd, if the method is defined. Otherwise, authenticationMethod is used. The daemon will not use the none authentication method.

The following example shows a section of a client profile in which the users will use sasl/digest-MD5 to authenticate to the directory server, but will use an SSL session to change their password.

serviceAuthenticationMethod=pam_ldap:sasl/digest-MD5
serviceAuthenticationMethod=passwd-cmd:tls:simple

Pluggable Authentication Methods

By using the PAM framework, you can choose among several authentication services, including pam_unix, pam_krb5, and pam_ldap.

If the per-user authentication method is used, pam_krb5, the strongest authentication service of the three methods listed above, must be enabled. See pam_krb5(5) and the System Administration Guide: Security Services.

The pam_krb5 authentication system may be used even if per-user authentication is not enabled. If proxy or anonymous credential levels are used to access directory server data then restricting access to directory data on a per-user basis is not possible.

Because of its increased flexibility, support of stronger authentication methods, and ability to use account management, the use of pam_ldap is recommended over the use of pam_unix when anonymous or proxy authentication methods are used.

pam_unix Service Modules

If you have not changed the pam.conf(4) file, pam_unix functionality is enabled by default.

The pam_unix module has been removed and is no longer supported with Solaris. A set of other service modules provides equivalent or greater functionality. Therefore, in this guide, pam_unix refers to the equivalent functionality, not to the pam_unix module itself.

Following is a list of the modules that provide the equivalent pam_unix functionality.

pam_unix follows the traditional model of UNIX authentication, as described in the following list.

1. The client retrieves the user's encrypted password from the name service.

2. The user is prompted for the user's password.

3. The user's password is encrypted.

4. The client compares the two encrypted passwords to determine whether the user should be authenticated.

Additionally, there are two restrictions when using pam_unix.

• The password must be stored in UNIX crypt format and not in any other encryption methods, including clear.

• The userPassword attribute must be readable by the name service.

  For example, if you set the credential level to anonymous, then anyone must be able to read the userPassword attribute. Similarly, if you set the credential level to proxy, then the proxy user must be able to read the userPassword attribute.

pam_unix is not compatible with the sasl authentication method digest-MD5, since Sun Java System Directory Server requires passwords to be stored in the clear in order to use digest-MD5. pam_unix requires the password be stored in crypt format.

Starting in the Solaris 10 10/09 release, pam_unix supports account management when the enableShadowUpdate switch is set to true. The controls for a remote LDAP user account are applied just as the controls are applied to a local user account that is defined in the passwd and shadow files. In enableShadowUpdate mode, for the LDAP account, the system updates and uses the shadow data on the LDAP server for password aging and account locking. Of course, the shadow data of the local account only applies to the local client system, whereas the shadow data of an LDAP user account applies to the user on all client systems.

Password history checking is only supported for the local client, not for an LDAP user account.

pam_krb5 Service Module

Refer to pam_krb5(5) and the System Administration Guide: Security Services.

pam_ldap Service Module

When implementing pam_ldap, the user binds to the LDAP server by using the authentication method defined in pam_ldap's serviceAuthenticationMethod parameter, if one exists. Otherwise, authenticationMethod is used.

If pam_ldap is able to bind to the server with the user's identity and supplied password, it authenticates the user.

Previously, if you enabled pam_ldap account management, all users needed to provide a login password for authentication any time they logged in to the system. Therefore, nonpassword-based logins using tools such as rsh, rlogin, or ssh would fail.

Now, however, pam_ldap(5), when used with Sun Java System Directory Servers DS5.2p4 and newer releases, enables users to log in with rsh, rlogin, rcp and ssh without giving a password.

pam_ldap(5) is now modified to perform account management and retrieve the account status of users without authenticating to Directory Server as the user logging in. The new control to this on Directory Server is 1.3.6.1.4.1.42.2.27.9.5.8, which is enabled by default.

To modify this control for other than default, add Access Control Instructions (ACI) on Directory Server:

dn: oid=1.3.6.1.4.1.42.2.27.9.5.8,cn=features,cn=config
objectClass: top
objectClass: directoryServerFeature
oid:1.3.6.1.4.1.42.2.27.9.5.8
cn:Password Policy Account Usable Request Control
aci: (targetattr != "aci")(version 3.0; acl "Account Usable"; 
     allow (read, search, compare, proxy)
     (groupdn = "ldap:///cn=Administrators,cn=config");)
creatorsName: cn=server,cn=plugins,cn=config
modifiersName: cn=server,cn=plugins,cn=config

pam_ldap does not read the userPassword attribute. Therefore, there is no need to grant access to read the userPassword attribute unless there are other clients using pam_unix. Also, pam_ldap does not support the none authentication method. Thus, you must define the serviceAuthenticationMethod or the authenticationMethod attributes so clients can use pam_ldap. See the pam_ldap(5) man page for more information.

If the simple authentication method is used, the userPassword attribute can be read on the wire by third parties.

See Example pam.conf File for pam_ldap.

The following table summarizes the main differences between pam_unix, pam_ldap, and pam_krb5.

PAM and Changing Passwords

Use the passwd command to change a password. If the enableShadowUpdate switch is not set to true, the userPassword attribute must be writable by the user. If the enableShadowUpdate switch is set to true, the admin credentials must be able to update the userPassword attribute. Remember that the serviceAuthenticationMethod for passwd-cmd overrides the authenticationMethod for this operation. Depending on the authentication method that is used, the current password might be unencrypted on the wire.

In the case of pam_unix, the new userPassword attribute is encrypted using UNIX crypt format and tagged before being written to LDAP. Therefore, the new password is encrypted on the wire, regardless of the authentication method used to bind to the server. See the pam_authtok_store(5) man page for more information.

If the enableShadowUpdate switch is set to true, pam_unix also updates the related shadow information when the user password is changed. pam_unix updates the same shadow fields in the local shadow files that pam_unix updates when the local user password is changed.

As of the Solaris 10 software release, pam_ldap no longer supports password update. The previously recommended use of pam_authtok_store with the server_policy option now replaces the pam_ldap password update capability. When you use pam_authtok_store, the new password is sent to the LDAP server in the clear. Therefore, to ensure privacy, use TLS. If TLS is not used, the new userPassword is subject to snooping. If you set an untagged password with Sun Java System Directory Server, the software encrypts the password by using the passwordStorageScheme attribute. For more information about the passwordStorageScheme, see the section on user account management in the Administration Guide for the version of Sun Java System Directory Server that you are using.

You need to consider the following configuration issues when setting the passwordStorageScheme attribute. If an NIS, NIS+, or another client using pam_unix is using LDAP as a repository, then passwordStorageScheme needs to be crypt. Also, if using pam_ldap with sasl/digest-MD5 with Sun Java System Directory Server, passwordStorageScheme must be set to clear.

Account Management

If you select pam_krb5 as your account and password management system, the Kerberos environment will manage all your account, password, account lockout, and other account management details. Refer to pam_krb5(5) and the System Administration Guide: Security Services.

If you do not use pam_krb5, then LDAP naming services can be configured to take advantage of the password and account lockout policy support in Sun Java System Directory Server. You can configure pam_ldap(5) to support user account management. passwd(1) enforces password syntax rules set by the Sun Java System Directory Server password policy, when used with the proper PAM configuration.

The following account management features are supported through pam_ldap(5). These features depend on Sun Java System Directory Server's password and account lockout policy configuration. You can enable as many or as few of the features as you want.

• Password aging and expiration notification

  Users must change their passwords according to a schedule. A password expires if it is not changed within the time configured. An expired password causes user authentication to fail.

  Users see a warning message whenever they log in within the expiration warning period. The message specifies the number of hours or days until the password expires.

• Password syntax checking

  New passwords must meet the minimum password length requirements. In addition, a password cannot match the value of the uid, cn, sn, or mail attributes in the user's directory entry.

• Password in history checking

  Users cannot reuse passwords. If a user attempts to change the password to one that was previously used, passwd(1) fails. LDAP administrators can configure the number of passwords kept in the server's history list.

• User account lockout

  A user account can be locked out after a given number of repeated authentication failures. A user can also be locked out if his account is inactivated by an administrator. Authentication will continue to fail until the account lockout time is passed or the administrator reactivates the account.

The preceding account management features only work with the Sun Java System Directory Server. For information about configuring the password and account lockout policy on the server, see the “User Account Management” chapter in the Administration Guide for the version of Sun Java System Directory Server that you are using. Also see Example pam_conf file for pam_ldap Configured for Account Management. Do not enable account management for proxy accounts.

Before configuring the password and account lockout policy on Sun Java System Directory Server, make sure all hosts use the “newest” LDAP client with pam_ldap account management.

In addition, make sure the clients have a properly configured pam.conf(4) file. Otherwise, LDAP naming services will not work when proxy or user passwords expire.

Previously, if you enabled pam_ldap account management, all users needed to provide a login password for authentication any time they logged in to the system. Therefore, nonpassword-based logins using tools such as rsh, rlogin, or ssh would fail.

Now, however, pam_ldap(5), when used with Sun Java System Directory Servers DS5.2p4 and newer releases, enables users to log in with rsh, rlogin, rcp and ssh without giving a password.

pam_ldap(5) is now modified to perform account management and retrieve the account status of users without authenticating to Directory Server as the user logging in. The new control to this on Directory Server is 1.3.6.1.4.1.42.2.27.9.5.8, which is enabled by default.

To modify this control for other than default, add Access Control Instructions (ACI) on Directory Server:

dn: oid=1.3.6.1.4.1.42.2.27.9.5.8,cn=features,cn=config
objectClass: top
objectClass: directoryServerFeature
oid:1.3.6.1.4.1.42.2.27.9.5.8
cn:Password Policy Account Usable Request Control
aci: (targetattr != "aci")(version 3.0; acl "Account Usable"; 
     allow (read, search, compare, proxy)
     (groupdn = "ldap:///cn=Administrators,cn=config");)
creatorsName: cn=server,cn=plugins,cn=config
modifiersName: cn=server,cn=plugins,cn=config

Account Management With pam_unix

Starting in the Solaris 10 10/09 release, the enableShadowUpdate switch is available. When enableShadowUpdate is set to true, LDAP offers the same functionality as the files naming service for account management.

If the enableShadowUpdate switch is set to true on the client, account management functionality that is available to local accounts is also available to LDAP accounts. Functionality includes password aging, account expiry and notification, failed login account locking, and so on. Also, the -dluNfnwx options to the passwd command are now supported in LDAP. Thus, the full functionality of the passwd command and the pam_unix* modules in the files naming service is supported in the LDAP naming service. The enableShadowUpdate switch provides a way to implement consistent account management for users who are defined in both the files and the LDAP scope.

To prevent users from modifying their own account management data and thereby circumventing password policy, the LDAP server is configured to prevent user write access to the user's own shadow data on the server. An administrator with admin credentials performs the shadow data updates for a client system. Such a configuration, however, conflicts with the pam_ldap module, which requires that passwords be modifiable by users. Therefore, account management by pam_ldap and pam_unix are incompatible.

Do not use both pam_ldap and pam_unix in the same LDAP naming domain. Either all clients use pam_ldap or all clients use pam_unix. This limitation might indicate that you need a dedicated LDAP server. For example, a web or email application might expect users to change their own password on the LDAP server.

The implementation of enableShadowUpdate also requires that the admin credential (adminDN plus adminPassword) be stored locally on every client. Even though adminPassword is encrypted and can only be read from the /var/ldap/ldap_client_cred file by the ldap_cachemgr daemon, special care must be taken to protect the admin credential. To protect the credential, make it different from the server's directory manager (cn=directory manager). Another protection would be to configure the serviceAuthenticationMethod with a value of tls:simple or better for the passwd-cmd service, so that the value of adminPassword is not sent in the clear and therefore becomes vulnerable to snooping.

Unlike using pam_ldap for account management, using pam_unix for account management does not require a change to the /etc/pam.conf file. The default /etc/pam.conf file is sufficient.