Centralized authentication using OpenLDAP/ja

このガイドはLDAPの基本を説明し,コンピュータのグループ間で認証を行なうためにどのようにOpenLDAPをセットアップするかを示します.

LDAPとは何ですか?
LDAPは "Lightweight Directory Access Protocol" の略です. X.500モデルに基づいておりその主要機能をカバーしていますが、X.500上では定義されているさらに難解な機能は除かれています. さて、このX.500とはどのようなものでしょうか？　なぜLDAPは規定されたのでしょうか？

X.500は、OSIの概念に基づくディレクトリサービスのモデルです. これには名前空間の定義と、ディレクトリのクエリと更新のためのプロトコルが含まれています. しかしながら、X.500は多くの場合過剰であることが判明しています. LDAPではX.500と同様にディレクトリ用のデータ・名前空間モデルとプロトコルを提供します. ただし、LDAPはTCP/IPスタックを介して直接実行されるように設計されています. よってLDAPはX.500の簡易版と言えます.

よくわかりません. ディレクトリとは何ですか？
ディレクトリは頻繁に参照され、あまり更新されない特殊なデータベースです. 一般的なデータベースとは異なり、トランザクションやロールバックはサポートされていません. ディレクトリは可用性と信頼性を高めるために容易に複製できます. ディレクトリが複製される時の一時的な不一致は、最終的に同期が取れる限り許容されます.

どのような情報構成ですか？
ディレクトリ内のすべての情報は階層構造になっています. さらに、ディレクトリにデータを入力する場合、データをツリー内に格納する方法を知っていなければなりません. 架空の会社とインターネットを模したツリーを見てみましょう:

Since you don't feed data to the database in this ascii-art like manner, every node of such a tree must be defined. To name such nodes, LDAP uses a naming scheme. Most LDAP distributions (including OpenLDAP) already contain quite a number of predefined (and general approved) schemas, such as the inetOrgPerson, or a frequently used schema to define users which Unix/Linux boxes can use, called posixAccount. Note there are GUI web based tools to make managing LDAP painless: see Working with OpenLDAP for an non-exhaustive list.

Interested users are encouraged to read the OpenLDAP Admin Guide.

So... what can it be used for?
LDAP can be used for various things. This document focuses on centralised user management, keeping all user accounts in a single LDAP location (which doesn't mean that it's housed on a single server, LDAP supports high availability and redundancy), yet other goals can be achieved using LDAP as well.


 * Public Key Infrastructure


 * Shared Calendar


 * Shared Addressbook


 * Storage for DHCP, DNS, ...


 * System Class Configuration Directives (keeping track of several server configurations)


 * Centralised Authentication (PosixAccount)



Common notes
The domain genfic.org is an example in this guide. You will of course want to change this. However, make sure that the top node is an official top level domain (net, com, cc, be, ...).

Let's first emerge OpenLDAP. Ensure the USE flags syslog, -minimal and tcpd are used.

OpenLDAP supports two authentication mechanisms:
 * 1) standard user-password (in LDAP terms user means binddn) named SIMPLE
 * 2) proxying authentication requests to SASL (Simple Authentication and Security Layer, see RFC4422 for details)

Although the OpenLDAP default is to use SASL, the initial version of this article used only password-based authentication. With the OLC add-on the article starts to describe the use of the simplest SASL mechanism called EXTERNAL, which relies on the system authentication.

OpenLDAP has a main user called "rootdn" (Root Distinguished Name), which is hardcoded in the application. Unlike the classic Unix root user, the rootdn user still needs to be assigned with proper permissions. The rootdn user may be used only in the context of the configuration, but it can also be used in the directory definition. In that case a user can authenticate himself as rootdn with either the configuration used password and the tree (directory-based) password.

User passwords (regardless if it is for rootdn users or others) for verification purposes can be stored as cleartext or hashed. Multiple different hash algorithms are available, but usage of weak algorythms (up to MD5) is not recommended. SHA is currently considered sufficiently cryptographically secure.

In the below command, a hashed value is created for a given password; the result of this command can be used in the configuration file, or in the internal directory definition of a user:

Legacy configuration (flat config slapd.conf)
Now edit the LDAP Server configuration in. The provided is from the original OpenLDAP source. Below is a sample configuration file one can use to replace it with to get things started.

For a more detailed analysis of the configuration file, we suggest that you work through the OpenLDAP Administrator's Guide, although may be enough.

If it does not start, the first thing you must do is to check the config file. You can do it with the following command.

Vary the debug level (the "-d 1" above) for more info. If all goes well you will see config file testing succeeded. If there's an error,  will list the line number to which it applies (of the  file).

By default writes the log events to the local4 syslog facility.

Migration from slapd.conf to OLC
If you want to be able to change OpenLDAP server's configuration, you must define at least  (or normally  ) access to.

The example below shows how to grant manage access on OLC (cn=config database) to the system administrator (root user) by adding the proper lines at the end of the file:

Then, we invoke the utility with the   and   options to convert the  file into a configuration directory.

Running this command will transfer and translate the configuration. After that you are expected to update the configuration using specially prepared ldif files. And only if you aren't enough familiar with them, you can first edit and after that re-translate the  into. Don't forget to check the directory's permissions.

For more instructions read the in-line comments of the generated files.

The below line will enable the configuration method.

/etc/conf.d/slapd

Finally, create the structure:

Initial setup with OLC
An initial configuration is shipped as a standard LDAP database dump, available as or.

It can be loaded (and only loaded, unlike ordinary LDAP databases) by the  utility:

If you use root account to do it, you must correct ownership of created files, as described below in migrate section.

If you need the right to change the configuration database, you must provide the proper permissions. The next example shows how these privileges are granted to the system user:

See for more details.

When using OLC, never manually edit the configuration files. The directory files can be used to check the consistency of the configuration through:

ディレクトリを管理する
slapd を始動させましょう:

Most users will also want the OpenLDAP daemon to start automatically:

It is now possible to use the directory server to authenticate users in apache/proftpd/qmail/samba.

The directory server can be managed with tools such as, and  from the Gentoo ebuild repository, or  from the poly-c overlay available through Layman or eselect repository.

Server management with OLC
Some examples of updates on the OLC-style configuration are mentioned below.

For instance, to change the location of the OLC configuration directory (needed after switching from a config file to config directory style):

{{Code|fix-configs.ldif| dn: cn=config changetype: modify delete: olcConfigFile

To change the log level used by the OpenLDAP instance:

{{Code|loglevel.ldif| dn: cn=config changetype: modify replace: olcLogLevel olcLogLevel: stats sync }}

In order to apply the changes, run the following command:

{{RootCmd|ldapmodify -Y EXTERNAL -H ldapi:/// -f loglevel.ldif}}

{{Warning|On restart, the init script performs a check of the updated configuration. The {{c|ldapmodify}} command used above blocks only fatal errors. To get info about non-fatal errors using OLC: {{RootCmd|slaptest -F /etc/openldap/slapd.d|output= 58b7d4c2 olcThreads: value #0: warning, threads=64 larger than twice the default (2*16=32); YMMV. config file testing succeeded }}}}

OpenLDAP logging
OpenLDAP produces numerous log events, which might not be obvious to interpret, but are necessary for debugging purposes.

As OpenLDAP by default writes the log events into the system log, it is advisable to reconfigure the system logger to direct OpenLDAP log events into a dedicated log file.

It is advisable to use the  log level in OpenLDAP standalone server and   in OpenLDAP cluster. In such case query results logs session-related information such as the following:

Most common errors in server log are :

Which means «invalid credentials» (i.e. wrong password).

And :

Which means «No such object». Usually this error appears when binddn (user) has no permissions on requested object. So either you try to do something wrong, or there is a mistake in your ACLs set.

Access management (ACLs)
The authorizations and access control mechanism used in OpenLDAP is described in the manual page. Its base syntax is as follows:

The following table shows the access levels available in OpenLDAP:

For details about the exact privilege settings, see the manual pages and official OpenLDAP documentation.

Config file
ACLs are parsed in the order they are set in the configuration, and are applied based on the specificity (meaning that, when an ACL rule is considered, the remainder of ACL rules is no longer checked). As such, more specific definitions should go first, before more generic ones are listed. For more information, see Access Control Evaluation.

For example:

Config directory
ACLs are parsed in the order they are set in the configuration, and are applied based on the specificity (meaning that, when an ACL rule is considered, the remainder of ACL rules is no longer checked). As such, more specific definitions should go first, before more generic ones are listed. This order, when using OLC, is handled through the  directives.

For example:

The following example inserts a new ACL on top, making the existing  entries to shift by one:

To delete an ACL:

High availability
A common high availability setup with OpenLDAP is to use replication of changes across multiple LDAP systems.

Replication within OpenLDAP is, in this guide, set up using a specific replication account which has read rights on the primary LDAP server and which pulls in changes from the primary LDAP server to the secondary.

This setup is then mirrored, allowing the secondary LDAP server to act as a primary. Thanks to OpenLDAP's internal structure, changes are not re-applied if they are already in the LDAP structure.

Setting up replication
To setup replication, first setup a second OpenLDAP server, similarly as above. However take care that, in the configuration file:


 * The sync replication provider is pointing to the other system


 * The serverID of each OpenLDAP system is different

Synchronisation account
Next, create the synchronisation account. We will create an LDIF file (the format used as data input for LDAP servers) and add it to each LDAP server:

Enabling syncprov overlay
Overlay can be linked statically and dynamically. When it is built dynamivally, you'll need to load module. For now in Gentoo it's usually built statically. To enshure type:

Load syncprov module (optional)
If you need to load syncprov module, you should use the following ldif file:

Setting up replication for database
Next step, mandatory for everybody, is to setup replication for database (must be done on both nodes):

Final configuration
Finally, you need to add replication's definition.

On node 1:

traditionally means the password string.

On node 2:

The only difference is in server's ident (rid) and provider uri.

If LDAP master (mirror node with initially loaded database) is unavailable (slapd daemon not started, or 389/tcp port is blocked by a packet filter) slapd daemon on secondary node fails to start with the following error message:

Almost certainly your database will not fit into default limits. So, you will need to encrease 's limits. For example:

Performance tuning
Default daemon settings significantly limitates LDAP server performance.

Sympthoms
When server load fits system limit client applications fails with different kind of timeout errors.

In server log this produces error messages like following:

Encreasing OS limits
First, you should read  system user limits:

The first parameter, you need to encrease, is the open files limit.

Maximum available value is described in Documentation/sysctl/fs.txt file of kernel documentation:

{{FileBox|filename=/usr/src/linux-4.9.95-gentoo/Documentation/sysctl/fs.txt|1=

=
=================================================

nr_open:

This denotes the maximum number of file-handles a process can allocate. Default value is 1024*1024 (1048576) which should be enough for most machines. Actual limit depends on RLIMIT_NOFILE resource limit.

=
================================================= }}

PAM system limits are stored in /etc/security/limits.conf file or, optionally, in /etc/security/limits.d/ directory. Daemons, started with init system use these parameters (see  for details), so you need just to put in the file:

And restart daemon.

The next limitation is    parameter. In runtime you could update it by:

After verifying new value do not forget to fix it:

And, possibly, some other application-specific parameters.

Configuring the OpenLDAP client tools
Edit the LDAP Client configuration file. This file is read by ldapsearch and other ldap command line tools.

You can test the running server with the following command:

If you receive an error, try adding  to increase the verbosity and solve the issue you have.

Client configuration for centralized authentication
There are numerous methods/tools that can be used for remote authentication. Some distributions also have their own easy to use configuration tool. Below there are some in no particular order. It is possible to combine local users and centrally authorized accounts at the same time. This is important because, for instance, if the LDAP server cannot be accessed one can still login as root.


 * SSSD (Single Sign-on Services Daemon). Its primary function is to provide access to identity and authentication remote resource through a common framework that can provide caching and offline support to the system. It provides PAM and NSS modules, and in the future will support D-Bus interfaces for extended user information. It also provides a better database to store local users as well as extended user data.


 * Use  to login to the LDAP server and authenticate. Passwords are not sent over the network in clear text.


 * NSLCD (Name Service Look up Daemon). Similar to SSSD, but older.


 * NSS (Name Service Switch) using the traditional  module to fetch password hashes over the network. To permit users to update their password this has to be combined with the   method.

The first two are demonstrated below with the minimum necessary configuration options to get working.

Client PAM configuration SSSD Method
Here is the more direct method. The three files that are required to be edited are mentioned below.

Add sss to the end as shown below to enable the lookup to be handed to the sssd system service. Once you have finished editing start the sssd daemon.

The last file is the most critical. Open an extra root terminal as a fallback before editing this. The lines that end with  have been added to enable remote authentication. Note the use of to support creating the user home directories.

Now try logging in from another box.

Client PAM configuration the pam_ldap module method
First, we will configure PAM to allow LDAP authorization. Install so that PAM supports LDAP authorization, and  so that your system can cope with LDAP servers for additional information (used by ).

The last file is the most critical. Open a few extra root terminals as a backup before editing this. The lines that end with  have been added to enable remote authentication.

/etc/pam.d/system-auth

Now change to read:

/etc/ldap.conf

Next, copy over the (OpenLDAP) file from the server to the client so the clients are aware of the LDAP environment:

Finally, configure your clients so that they check the LDAP for system accounts:

/etc/nsswitch.conf

If you noticed one of the lines you pasted into your was commented out (the   line): you don't need it unless you want to change a user's password as superuser. In this case you need to echo the root password to in plaintext. This is DANGEROUS and should be chmoded to 600. What you might want to do is keep that file blank and when you need to change someone's password that's both in the LDAP and, put the pass in there for 10 seconds while changing the users password and remove it when done.

Convert file userbase to LDAP
Configuring OpenLDAP for centralized administration and management of common Linux/Unix items isn't easy, but thanks to some tools and scripts available on the Internet, migrating a system from a single-system administrative point-of-view towards an OpenLDAP-based, centralized managed system isn't hard either.

Go to https://www.padl.com/OSS/MigrationTools.html and fetch the scripts there. You'll need the migration tools and the script.

Next, extract the tools and copy the script inside the extracted location:

The next step now is to migrate the information of your system to OpenLDAP. The script will do this for you, after you have provided it with the information regarding your LDAP structure and environment.

At the time of writing, the tools require the following input:

The tool will also ask you which accounts and settings you want to migrate.

謝辞
このガイドの為に環境を貸してくれた Matt heler に感謝します. また、@ irc.freenode.net の #ldap にいる方々に感謝します.