ms-active-directory 1.4.0

Python library for integrating with Microsoft Active Directory

ms_active_directory - A Library for Integrating with Microsoft Active Directory

This is a library for integrating with Microsoft Active Directory domains.

It supports a variety of common, critical functionality for integration of computers into a domain, including the ability to discover domain resources, optimize communication for speed, join a computer to the domain, and look up information about users and groups in the domain.

It primarily builds on the LDAP protocol, and supports LDAP over TLS with channel bindings, and all LDAP basic, NTLM, and SASL authentication mechanisms supported by the ldap3 python library.

Author: Azaria Zornberg
Email: [e-mail]

Key Features

1. Joining a computer to a domain, either by creating a new computer account or taking over an existing account.
2. Discovering domain resources and properties, optimizing domain communication for the network
3. Finding users and groups based on a variety of properties (e.g. name, SID, user-specified properties)
4. Querying information about users, groups, and generic objects
5. Looking up user and group memberships
6. Adding and removing users and groups to and from other groups
7. Account management functionality, such as password changes/resets, enabling, disabling, and unlocking accounts
8. Looking up information about the permissions set on a user, group, or generic object
9. Adding permissions to the security descriptor for a user, group, or generic object

Examples

Discovering a domain

The library supports discovering LDAP and Kerberos servers within a domain using special DNS entries defined for Active Directory.

Smart Defaults

By default, it will use the system DNS configuration, find LDAP servers that support TLS, and sort LDAP and Kerberos servers by the RTT to communicate with them.

from ms_active_directory import ADDomain

example_domain_dns_name = 'example.com'
domain = ADDomain(example_domain_dns_name)
ldap_servers = domain.get_ldap_uris()
kerberos_servers = domain.get_kerberos_uris()

# re-discover servers in dns and sort them by RTT again at a later time to pick up changes
domain.refresh_ldap_server_discovery()
domain.refresh_kerberos_server_discovery()

Site Awareness and Flexible DNS

The library also supports site awareness, which will result in only discovering servers within a specified Active Directory Site. You can also specify alternative DNS nameservers to use instead of the system ones.

from ms_active_directory import ADDomain

example_domain_dns_name = 'example.com'
site_name = 'us-eastern-datacenter'
domain = ADDomain(example_domain_dns_name, site=site_name,
                  dns_nameservers=['eastern-private-dns-01.local'])

Network Multi-Tenancy and Security Support

You can also specify exactly which LDAP or Kerberos servers should be used, and skip discovery. Additional configurations are available such as configuring the CA file path to use for trust, and the source IP to use for outbound traffic to the domain, which is helpful when there are firewall rules in place, or when a machine has both private and public IP addresses.

from ms_active_directory import ADDomain

example_domain_dns_name = 'example.com'
local_machine_ip = '10.251.12.1'
local_ldap_ip = '10.251.12.30'
public_machine_ip = '194.32.21.30'
# the servers that live on the public internet use well-known public
# CAs for trust, but we have a local CA for the private network servers
private_securing_cas = '/etc/internal-ca.cert'

# set up an object for the local domain in the same network as this machine,
# but also have an instance that can be used to make instances to reach out
# to the rest of the domain outside of the local private network
local_domain = ADDomain(example_domain_dns_name, ldap_servers_or_uris=[local_ldap_ip],
                        source_ip=local_ldap_ip, ca_certificates_file_path=private_securing_cas)
global_domain = ADDomain(example_domain_dns_name, source_ip=public_machine_ip)

Establishing a session with a domain

You can establish a session with the AD Domain on behalf of either a user or computer.

Broadly, any keyword arguments that would normally be supported when creating a Connection with the ldap3 library are supported when creating a session, allowing for flexibility while still providing an "it just works" option for most users.

Support for Computer Authentication

Computers default to using Kerberos SASL authentication, as SIMPLE authentication is not support for computers with Active Directory. To use kerberos, either gssapi or winkerberos must be installed.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')

# when using kerberos auth, the default is to use the kerberos
# credential cache on the machine, so no password is needed
computer_name = 'machine01'
session1 = domain.create_session_as_computer(computer_name)

# but you can pass sasl credentials, and if you use gssapi you can
# specify a username and password
# see the ldap3 documentation for details on SASL credentials and other
# connection options
other_name = 'other-machine-identity'
password = 'password01'
session2 = domain.create_session_as_computer(other_name, sasl_credentials=('', other_name, password))

You can also use other authentication mechanisms like NTLM.

from ldap3 import NTLM
from ms_active_directory import ADDomain
domain = ADDomain('example.com')

ntlm_name = 'EXAMPLE.COM\computer01'
password = 'password1'
session = domain.create_session_as_computer(ntlm_name, password, authentication_mechanism=NTLM)

Support for User Authentication

You can authenticate as a user by using simple binds, or by using SASL mechanisms or NTLM as computers do. The default for users is simple binds.

from ldap3 import NTLM
from ms_active_directory import ADDomain
domain = ADDomain('example.com')

session = domain.create_session_as_user('[e-mail]', 'password')
ntlm_session =  domain.create_session_as_user('[e-mail]', 'password',
                                              authentication_mechanism=NTLM)

Discovering additional domain resources

The library supports discovering a wide variety of information about the domain beyond the basics needed to communicate with it. This discovery doesn't require you to know any niche information about Active Directory.

Discoverable resources include but are not limited to:

• Supported SASL mechanisms, which is important for authentication
• The current domain time, which is important for NTP synchronization
• Domain Functional Level, which governs things like support encryption types
• DNS servers
• Issuing certificates for CAs in the domain

Finding supported SASL mechanisms

Discovering SASL mechanisms can be done without needing to create a session with a domain, as it's needed before authentication in many cases.

from ms_active_directory import ADDomain
domain = ADDomain('example.com') 

# might print "['EXTERNAL', 'DIGEST-MD5']"
print(domain.find_supported_sasl_mechanisms())

Finding the current domain time

Discovering the domain time can be done without needing to create a session with a domain, as time synchronization is necessary for kerberos authentication to succeed and can impact TLS negotiation as well.

from ms_active_directory import ADDomain
domain = ADDomain('example.com') 

# returns a python datetime object in utc time
curr_time = domain.find_current_time()

# allowed drift defaults to 5 minutes which is the kerberos standard,
# but we can use a shorter window to detect drift before it causes an
# outage. this returns a boolean
synced = domain.is_close_in_time_to_localhost(allowed_drift_seconds=60)

Finding the domain functional level

Discovering the domain time can be done without needing to create a session with a domain, as it can inform us as to what encryption types and TLS versions/ciphers will be supported by the domain.

from ms_active_directory import ADDomain
domain = ADDomain('example.com') 

# find_functional_level returns an enum indicating the level.
# decision making based on level should be done based on the
# needs of your application
print(domain.find_functional_level())

Finding DNS servers

Discovering DNS servers requires an authenticated session with the domain, as searching the records within the domain for computers that run a DNS service is privileged.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')

session = domain.create_session_as_user('[e-mail]', 'password')
# returns a map that maps server hostnames -> ip addresses, where
# the hostnames are computers running dns services
dns_map = session.find_dns_servers_for_domain()
ip_addrs = dns_map.values()
hostnames = dns_map.keys()

Finding CA certificates

Discovering DNS servers requires an authenticated session with the domain, as searching the records within the domain for records that are indicated as being certificate authorities is privileged.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')

session = domain.create_session_as_user('[e-mail]', 'password')
# returns a list of PEM-formatted strings representing the signing certificates
# of all certificate authorities in the domain
pem_certs = session.find_certificate_authorities_for_domain()

# you can also get the certificates in DER format, which might be
# preferred on windows
der_certs = session.find_certificate_authorities_for_domain(pem_format=False)

Joining an Active Directory domain

The action of joining a computer to a domain is not a well-defined operation, and so the exact mechanics of how you utilize the domain joining functionality and how its outputs are integrated with the rest of your system will vary depending on your use case.

This will try to cover some common examples.

Join the domain with default configurations for everything

The default behavior requires only the domain name and the credentials of a user with sufficient administrative rights to create computers within the domain.

from ms_active_directory import join_ad_domain

comp = join_ad_domain('example.com', '[e-mail]', 'example-password')

The join_ad_domain function returns an ADComputer object with many helpful functions describing properties of the created computer.

This will use the local hostname of the machine running this code as the computer name. It will create the computer in AD's default Computers container.

It enables AES256-SHA1 as an encryption type for both receiving and initiating kerberos contexts, and it configures <local hostname>.<domain dns name> as the hostname of the computer in AD and registers the default HOST service.

It then writes kerberos keys for the new computer account to /etc/krb5.keytab, which is the default location for kerberos keytabs.

This all enables the account to be used for authenticating with other domain resources as a client over protocols like SMB and LDAP using kerberos, as well as receiving incoming kerberos authentication as a server for things like SSH. This is because the HOST service encapsulates many standard services in the domain.

However, it is still up to the caller to do things like configure sshd to utilize the keytab.

Join the domain with customization of the account for security reasons

A number of customizations exist for security reasons.

You can change things like the encryption types enabled on the account to support older clients. You can also change location where the account is created when joining a domain in order to use a less privileged user for the act of joining.

You can also set the computer name if you have a desired naming scheme. This will impact the hostnames configured in the domain for the computer.

from ms_active_directory import join_ad_domain, ADEncryptionType

domain = 'example.com'
less_privileged_user = '[e-mail]'
password = 'password2'
# ldap-style relative distinguished name of a location
less_privileged_loc = 'OU=service-machines,OU=ops'
computer_name = 'workstation10'

legacy_enc_type = ADEncryptionType.RC4_HMAC
new_enc_type = ADEncryptionType.AES256_CTS_HMAC_SHA1_96

comp = join_ad_domain(domain, less_privileged_user, password, computer_name=computer_name,
                      computer_location=password, computer_encryption_types=[legacy_enc_type, new_enc_type])

You can also manually set the computer password. The default is to generate a random 120 character password, but if you want to share this computer across services, and some cannot interact with the generated kerberos keys, then you may wish to set a password manually.

You can also change where the kerberos keys are written to.

from ms_active_directory import join_ad_domain

domain = 'example.com'
user = '[e-mail]'
password = 'password2'
kerberos_key_location = '/usr/shared/keys/workstation-key.keytab'
computer_name = 'workstation10'
computer_password = 'workstation-shared-pw'

comp = join_ad_domain(domain, user, password, computer_key_file_path=kerberos_key_location,
                      computer_name=computer_name, computer_password=computer_password)

Join the domain with different network or service settings

You can configure different hostnames for your computer when joining the domain. This is useful when you have multiple different hostnames for a single device, or want to use a computer name that doesn't match your network name.

You can also configure services in order to restrict or broaden what is supported by the computer when acting as a server (e.g. you can add nfs if the machine will be an nfs server).

Joining will fail if another computer in the domain is using the services you specify on any of the hostnames you specify in order to avoid conflicts that cause undefined behavior.

from ms_active_directory import join_ad_domain

domain = 'example.com'
user = '[e-mail]'
password = 'password2'

services = ['HOST', 'nfs', 'cifs', 'HTTP']
computer_name = 'workstation10'
computer_host1 = 'central-mount-point.example.com'
computer_host2 = 'example-web-server.example.com'
comp = join_ad_domain(domain, user, password, computer_name=computer_name,
                      computer_hostnames=[computer_host1, computer_host2],
                      computer_services=services)

Join using a domain object

You can use an ADDomain object to join the domain as well, using a join function. This allows you to combine all of the functionality mentioned earlier around site-awareness, server preferences, TLS settings, and network multi-tenancy with the domain joining functionality mentioned in this section.

The parameters are all the same, except the domain need not be provided when using an ADDomain object, so it just adds more functionality in exchange for a slightly less simple workflow.

from ms_active_directory import ADDomain

domain = ADDomain('example.com', site='us-eastern-dc',
                  source_ip='10.25.21.30', dns_nameservers=['10.25.21.20'])

user = '[e-mail]'
password = 'password2'
less_privileged_loc = 'OU=service-machines,OU=ops'
services = ['HOST', 'nfs', 'cifs', 'HTTP']
computer_name = 'workstation10'

comp = domain.join(user, password, computer_hostnames=[computer_host1, computer_host2],
                   computer_services=services, computer_location=less_privileged_loc)

Join the domain by taking over an existing account

For some setups, accounts may be pre-created and then taken over by the computers that will use them.

This can be done in order to greatly restrict the permissions of the user that is used for joining, as they only need RESET PASSWORD permissions on the computer account, or CHANGE PASSWORD if the current computer password is provided.

from ms_active_directory import ADDomain, join_ad_domain_by_taking_over_existing_computer

domain_dns_name = 'example.com'
site = 'us-eastern-dc'
existing_computer_name = 'precreated-comp'
user = '[e-mail]'
password = 'password2'

computer_obj = join_ad_domain_by_taking_over_existing_computer(domain_dns_name, user, password,
                                                               ad_site=site, computer_name=existing_computer_name)
                                                               
# or use a domain object to use various power-user domain features
domain = ADDomain(domain_dns_name, site=site,
                  source_ip='10.25.21.30', dns_nameservers=['10.25.21.20'])
domain.join_by_taking_over_existing_computer(user, password, computer_name=existing_computer_name)

Managing users and groups

The library provides a number of different functions for finding users and groups by different identifiers, and for querying information about them. It also has functions for checking their memberships and adding or removing users and groups to or from groups.

Looking up users, groups, and information about them

Users and groups can both be looked up by one of:

• sAMAccountName
• distinguished name
• common name
• a generic "name" that will attempt the above 3
• an attribute

Look up by sAMAccountName

A sAMAccountName is unique within a domain, and so looking up users or groups by sAMAccountName returns a single result. sAMAccountName was a user's windows logon name in older versions of windows, and may be referred to as such in some documentation.

When looking up users and groups, you can also query for additional information about them by specifying a list of LDAP attributes.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user = session.find_user_by_sam_name('user1', ['employeeID'])
group = session.find_group_by_sam_name('group1', ['gidNumber'])
# users and groups support a generic "get" for any attributes queried
print(user.get('employeeID'))
print(group.get('gidNumber'))

Look up by distinguished name

A distinguished name is unique within a forest, and so looking up users or groups by it returns a single result. A distinguished name should not be escaped when provided to the search function.

When looking up users and groups, you can also query for additional information about them by specifying a list of LDAP attributes.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user_dn = 'CN=user one,CN=Users,DC=example,DC=com'
user = session.find_user_by_distinguished_name(user_dn, ['employeeID'])
group_dn = 'CN=group one,OU=employee-groups,DC=example,DC=com'
group = session.find_group_by_distinguished_name(group_dn, ['gidNumber'])
# users and groups support a generic "get" for any attributes queried
print(user.get('employeeID'))
print(group.get('gidNumber'))

Look up by common name

A common name is not unique within a domain, and so looking up users or groups by it returns a list of results, which may have 0 or more entries.

When looking up users and groups, you can also query for additional information about them by specifying a list of LDAP attributes.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user_cn = 'John Doe'
users = session.find_users_by_common_name(user_cn, ['employeeID'])
group_dn = 'operations managers'
groups = session.find_groups_by_common_name(group_dn, ['gidNumber'])
# users and groups support a generic "get" for any attributes queried
for user in users:
    print(user.get('employeeID'))
for group in groups:
    print(group.get('gidNumber'))

Look up by generic name

You can also query by a generic "name" and the library will attempt to find a unique user or group with that name. The library will either lookup by DN or will attempt sAMAccountName and common name lookups depending on the name format.

If more than one result is found by common name and no result is found by sAMAccountName then this will produce an error.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user_name = 'John Doe'
user = session.find_user_by_name(user_name, ['employeeID'])
group_name = 'operations managers'
groups = session.find_groups_by_name(group_name, ['gidNumber'])
# users and groups support a generic "get" for any attributes queried
print(user.get('employeeID'))
print(group.get('gidNumber'))

Look up by attribute

You can also query for users or groups that possess a certain value for a specified attribute. This can produce any number of results so a list is returned.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

desired_employee_type = 'temporary'
users = session.find_users_by_attribute('employeeType', desired_employee_type, ['employeeID'])
desired_group_manager = 'Alice P Hacker'
groups = session.find_groups_by_attribute('managedBy', desired_group_manager, ['gidNumber'])

# users and groups support a generic "get" for any attributes queried
for user in users:
    print(user.distinguished_name)
    print(user.get('employeeID'))
for group in groups:
    print(group.distinguished_name)
    print(group.get('gidNumber'))

Querying user and group membership

You can also look up the groups that a user belongs to, or the groups that a group belongs to. Active Directory supports nested groups, which is why there's both user->groups and group->groups mapping capability.

When querying the membership information for users or groups, the input type for any user or group must either be a string name identifying the user or group as described in the prior section, or must be an ADUser or ADGroup object returned by one of the functions described in the prior section.

Similarly to looking up users and groups, you can query for attributes of the parent groups by providing a list of LDAP attributes to look up for them.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user_sam_account_name = 'user-sam-1'
user_dn = 'CN=user sam 1,CN=users,DC=example,DC=com'
user_cn = 'user same 1'

desired_group_attrs = ['gidNumber', 'managedBy']
# all 3 of these do the same thing, and internally map the different
# name types to a user object
groups_res1 = session.find_groups_for_user(user_sam_account_name, desired_group_attrs)
groups_res2 = session.find_groups_for_user(user_dn, desired_group_attrs)
groups_res3 = session.find_groups_for_user(user_cn, desired_group_attrs)

# you can also directly use a user object to query groups
user_obj = session.find_user_by_name(user_sam_account_name)
groups_res4 = session.find_groups_for_user(user_obj, desired_group_attrs)

# you can also look up the parents of groups in the same way
example_group_obj = groups_res4[0]
example_group_dn = example_group_obj.distinguished_name

# these both work. sAMAccountName could also be used, etc.
second_level_groups_res1 = session.find_groups_for_group(example_group_obj, desired_group_attrs)
second_level_groups_res2 = session.find_groups_for_group(example_group_dn, desired_group_attrs)

You can also query users->groups and groups->groups to find the memberships of multiple users and groups, and the library will make a minimal number of queries to determine membership; it will be more efficient that doing a user->groups for each user (or similar for groups). The result will be a map that maps the input users or groups to lists of parent groups.

The input lists' elements must be the same format as what's provided when looking up group memberships for a single user or group.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user1_name = 'user1'
user2_name = 'user2'
users = [user1_name, user2_name]
desired_group_attrs = ['gidNumber', 'managedBy']

user_group_map = session.find_groups_for_users(users, desired_group_attrs)
# the dictionary result keys are the users from the input
user1_groups = user_group_map[user1_name]
user2_groups = user_group_map[user2_name]

# you can use the groups->groups mapping functionality to enumerate the
# full tree of a users' group memberships (or a groups' group memberships)
user1_second_level_groups_map = session.find_groups_for_groups(user1_groups, desired_group_attrs)
all_second_level_groups = []
for group_list in user1_second_level_groups_map.values():
    for group in group_list:
        if group not in all_second_level_groups:
            all_second_level_groups.append(group)
all_user1_groups_in_2_levels = user1_groups + all_second_level_groups

Adding users to groups

You can add users to groups by specifying a list of ADUser objects or string names of AD users to be added to the groups, and a list of ADGroup objects or string names of AD groups to add the users to.

If string names are specified, they'll be mapped to users/groups using the functions discussed in the prior sections.

If a user is already in a group, this is idempotent and will not re-add them.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user1_name = 'user1'
user2_name = 'user2'
group1_name = 'target-group1'
group2_name = 'target-group2'

session.add_users_to_groups([user1_name, user2_name],
                            [group1_name, group2_name])

By default, if we fail to add users to one of the groups specified, we'll attempt to rollback and remove users from any groups they were added to. You can choose to forgo this and a list of groups that users were successfully added to will be returned instead.

from ms_active_directory import ADDomain
domain = ADDomain('example.com')
session = domain.create_session_as_user('[e-mail]', 'password')

user1_name = 'user1'
user2_name = 'user2'
group1_name = 'target-group1'
group2_name = 'target-group2'
privileged_group = 'group-that-will-fail'

succeeeded = session.add_users_to_groups([user1_name, user2_name],
                                         [group1_name, group2_name, privileged_group],
                                         stop_and_rollback_on_error=False)
# this will print "['target-group1', 'target-group2']" assuming that
# adding users to 'group-that-will-fail' failed
print(succeeeded)                                 

Adding groups to groups

Adding groups to other groups works exactly the same way as adding users to groups, but the function is called add_groups_to_groups and both inputs are lists of groups.

Removing users or groups from groups

Removing users or groups from groups works identically to adding users or groups to groups, including input format, idempotency, and rollback functionality. The only difference is that the functions are called remove_users_from_groups and remove_groups_from_groups instead.