- Common Problems & Workflows
-
Could not authenticate you from Ldapmain because "Unknown provider"
- Expired license causes errors with multiple LDAP servers
- Debugging Tools
Troubleshooting LDAP
If you are an administrator, use the following information to troubleshoot LDAP.
Common Problems & Workflows
Connection
Connection refused
If you’re getting Connection Refused
error messages when attempting to
connect to the LDAP server, review the LDAP port
and encryption
settings
used by GitLab. Common combinations are encryption: 'plain'
and port: 389
,
or encryption: 'simple_tls'
and port: 636
.
Connection times out
If GitLab cannot reach your LDAP endpoint, you see a message like this:
Could not authenticate you from Ldapmain because "Connection timed out - user specified timeout".
If your configured LDAP provider and/or endpoint is offline or otherwise unreachable by GitLab, no LDAP user is able to authenticate and sign-in. GitLab does not cache or store credentials for LDAP users to provide authentication during an LDAP outage.
Contact your LDAP provider or administrator if you are seeing this error.
Referral error
If you see LDAP search error: Referral
in the logs, or when troubleshooting
LDAP Group Sync, this error may indicate a configuration problem. The LDAP
configuration /etc/gitlab/gitlab.rb
(Omnibus) or config/gitlab.yml
(source)
is in YAML format and is sensitive to indentation. Check that group_base
and
admin_group
configuration keys are indented 2 spaces past the server
identifier. The default identifier is main
and an example snippet looks like
the following:
main: # 'main' is the GitLab 'provider ID' of this LDAP server
label: 'LDAP'
host: 'ldap.example.com'
...
group_base: 'cn=my_group,ou=groups,dc=example,dc=com'
admin_group: 'my_admin_group'
Query LDAP
The following allows you to perform a search in LDAP using the rails console.
Depending on what you’re trying to do, it may make more sense to query a user
or a group directly, or even use ldapsearch
instead.
adapter = Gitlab::Auth::Ldap::Adapter.new('ldapmain')
options = {
# :base is required
# use .base or .group_base
base: adapter.config.group_base,
# :filter is optional
# 'cn' looks for all "cn"s under :base
# '*' is the search string - here, it's a wildcard
filter: Net::LDAP::Filter.eq('cn', '*'),
# :attributes is optional
# the attributes we want to get returned
attributes: %w(dn cn memberuid member submember uniquemember memberof)
}
adapter.ldap_search(options)
When using OIDs in the filter, replace Net::LDAP::Filter.eq
with Net::LDAP::Filter.construct
:
adapter = Gitlab::Auth::Ldap::Adapter.new('ldapmain')
options = {
# :base is required
# use .base or .group_base
base: adapter.config.base,
# :filter is optional
# This filter includes OID 1.2.840.113556.1.4.1941
# It will search for all direct and nested members of the group gitlab_grp in the LDAP directory
filter: Net::LDAP::Filter.construct("(memberOf:1.2.840.113556.1.4.1941:=CN=gitlab_grp,DC=example,DC=com)"),
# :attributes is optional
# the attributes we want to get returned
attributes: %w(dn cn memberuid member submember uniquemember memberof)
}
adapter.ldap_search(options)
For examples of how this is run,
review the Adapter
module.
User sign-ins
No users are found
If you’ve confirmed that a connection to LDAP can be established but GitLab doesn’t show you LDAP users in the output, one of the following is most likely true:
- The
bind_dn
user doesn’t have enough permissions to traverse the user tree. - The users don’t fall under the configured
base
. - The configured
user_filter
blocks access to the users.
In this case, you con confirm which of the above is true using
ldapsearch with the existing LDAP configuration in your
/etc/gitlab/gitlab.rb
.
Users cannot sign-in
A user can have trouble signing in for any number of reasons. To get started, here are some questions to ask yourself:
- Does the user fall under the configured
base
in LDAP? The user must fall under thisbase
to sign in. - Does the user pass through the configured
user_filter
? If one is not configured, this question can be ignored. If it is, then the user must also pass through this filter to be allowed to sign in.- Refer to our documentation on debugging the
user_filter
.
- Refer to our documentation on debugging the
If the above are both okay, the next place to look for the problem is the logs themselves while reproducing the issue.
- Ask the user to sign in and let it fail.
- Look through the output for any errors or other messages about the sign-in. You may see one of the other error messages on this page, in which case that section can help resolve the issue.
If the logs don’t lead to the root of the problem, use the rails console to query this user to see if GitLab can read this user on the LDAP server.
It can also be helpful to debug a user sync to investigate further.
Invalid credentials on sign-in
If that the sign-in credentials used are accurate on LDAP, ensure the following are true for the user in question:
- Make sure the user you are binding with has enough permissions to read the user’s tree and traverse it.
- Check that the
user_filter
is not blocking otherwise valid users. - Run an LDAP check command to make sure that the LDAP settings are correct and GitLab can see your users.
Access denied for your LDAP account
There is a bug that
may affect users with Auditor level access. When
downgrading from Premium/Ultimate, Auditor users who try to sign in
may see the following message: Access denied for your LDAP account
.
We have a workaround, based on toggling the access level of affected users:
- As an administrator, on the top bar, select Main menu > Admin.
- On the left sidebar, select Overview > Users.
- Select the name of the affected user.
- In the upper right, select Edit.
- Change the user’s access level from
Regular
toAdministrator
(or vice versa). - At the bottom of the page, select Save changes.
- In the upper right, select Edit again.
- Restore the user’s original access level (
Regular
orAdministrator
) and select Save changes again.
The user should now be able to sign in.
Email has already been taken
A user tries to sign in with the correct LDAP credentials, is denied access, and the production.log shows an error that looks like this:
(LDAP) Error saving user <USER DN> (email@example.com): ["Email has already been taken"]
This error is referring to the email address in LDAP, email@example.com
. Email
addresses must be unique in GitLab and LDAP links to a user’s primary email (as opposed
to any of their possibly-numerous secondary emails). Another user (or even the
same user) has the email email@example.com
set as a secondary email, which
is throwing this error.
We can check where this conflicting email address is coming from using the rails console. In the console, run the following:
# This searches for an email among the primary AND secondary emails
user = User.find_by_any_email('email@example.com')
user.username
This shows you which user has this email address. One of two steps must be taken here:
- To create a new GitLab user/username for this user when signing in with LDAP, remove the secondary email to remove the conflict.
- To use an existing GitLab user/username for this user to use with LDAP, remove this email as a secondary email and make it a primary one so GitLab associates this profile to the LDAP identity.
The user can do either of these steps in their profile or an administrator can do it.
Projects limit errors
The following errors indicate that a limit or restriction is activated, but an associated data field contains no data:
-
Projects limit can't be blank
. -
Projects limit is not a number
.
To resolve this:
- On the top bar, select Main menu > Admin.
- On the left sidebar, go to Settings > General.
- Expand both of the following:
- Account and limit.
- Sign-up restrictions.
- Check, for example, the Default projects limit or Allowed domains for sign-ups fields and ensure that a relevant value is configured.
Debug LDAP user filter
ldapsearch
allows you to test your configured
user filter
to confirm that it returns the users you expect it to return.
ldapsearch -H ldaps://$host:$port -D "$bind_dn" -y bind_dn_password.txt -b "$base" "$user_filter" sAMAccountName
- Variables beginning with a
$
refer to a variable from the LDAP section of your configuration file. - Replace
ldaps://
withldap://
if you are using the plain authentication method. Port389
is the defaultldap://
port and636
is the defaultldaps://
port. - We are assuming the password for the
bind_dn
user is inbind_dn_password.txt
.
Sync all users
The output from a manual user sync can show you what happens when GitLab tries to sync its users against LDAP. Enter the rails console and then run:
Rails.logger.level = Logger::DEBUG
LdapSyncWorker.new.perform
Next, learn how to read the output.
Example console output after a user sync
The output from a manual user sync is very verbose, and a single user’s successful sync can look like this:
Syncing user John, email@example.com
Identity Load (0.9ms) SELECT "identities".* FROM "identities" WHERE "identities"."user_id" = 20 AND (provider LIKE 'ldap%') LIMIT 1
Instantiating Gitlab::Auth::Ldap::Person with LDIF:
dn: cn=John Smith,ou=people,dc=example,dc=com
cn: John Smith
mail: email@example.com
memberof: cn=admin_staff,ou=people,dc=example,dc=com
uid: John
UserSyncedAttributesMetadata Load (0.9ms) SELECT "user_synced_attributes_metadata".* FROM "user_synced_attributes_metadata" WHERE "user_synced_attributes_metadata"."user_id" = 20 LIMIT 1
(0.3ms) BEGIN
Namespace Load (1.0ms) SELECT "namespaces".* FROM "namespaces" WHERE "namespaces"."owner_id" = 20 AND "namespaces"."type" IS NULL LIMIT 1
Route Load (0.8ms) SELECT "routes".* FROM "routes" WHERE "routes"."source_id" = 27 AND "routes"."source_type" = 'Namespace' LIMIT 1
Ci::Runner Load (1.1ms) SELECT "ci_runners".* FROM "ci_runners" INNER JOIN "ci_runner_namespaces" ON "ci_runners"."id" = "ci_runner_namespaces"."runner_id" WHERE "ci_runner_namespaces"."namespace_id" = 27
(0.7ms) COMMIT
(0.4ms) BEGIN
Route Load (0.8ms) SELECT "routes".* FROM "routes" WHERE (LOWER("routes"."path") = LOWER('John'))
Namespace Load (1.0ms) SELECT "namespaces".* FROM "namespaces" WHERE "namespaces"."id" = 27 LIMIT 1
Route Exists (0.9ms) SELECT 1 AS one FROM "routes" WHERE LOWER("routes"."path") = LOWER('John') AND "routes"."id" != 50 LIMIT 1
User Update (1.1ms) UPDATE "users" SET "updated_at" = '2019-10-17 14:40:59.751685', "last_credential_check_at" = '2019-10-17 14:40:59.738714' WHERE "users"."id" = 20
There’s a lot here, so let’s go over what could be helpful when debugging.
First, GitLab looks for all users that have previously signed in with LDAP and iterate on them. Each user’s sync starts with the following line that contains the user’s username and email, as they exist in GitLab now:
Syncing user John, email@example.com
If you don’t find a particular user’s GitLab email in the output, then that user hasn’t signed in with LDAP yet.
Next, GitLab searches its identities
table for the existing
link between this user and the configured LDAP providers:
Identity Load (0.9ms) SELECT "identities".* FROM "identities" WHERE "identities"."user_id" = 20 AND (provider LIKE 'ldap%') LIMIT 1
The identity object has the DN that GitLab uses to look for the user in LDAP. If the DN isn’t found, the email is used instead. We can see that this user is found in LDAP:
Instantiating Gitlab::Auth::Ldap::Person with LDIF:
dn: cn=John Smith,ou=people,dc=example,dc=com
cn: John Smith
mail: email@example.com
memberof: cn=admin_staff,ou=people,dc=example,dc=com
uid: John
If the user wasn’t found in LDAP with either the DN or email, you may see the following message instead:
LDAP search error: No Such Object
…in which case the user is blocked:
User Update (0.4ms) UPDATE "users" SET "state" = $1, "updated_at" = $2 WHERE "users"."id" = $3 [["state", "ldap_blocked"], ["updated_at", "2019-10-18 15:46:22.902177"], ["id", 20]]
After the user is found in LDAP, the rest of the output updates the GitLab database with any changes.
Query a user in LDAP
This tests that GitLab can reach out to LDAP and read a particular user. It can expose potential errors connecting to and/or querying LDAP that may seem to fail silently in the GitLab UI.
Rails.logger.level = Logger::DEBUG
adapter = Gitlab::Auth::Ldap::Adapter.new('ldapmain') # If `main` is the LDAP provider
Gitlab::Auth::Ldap::Person.find_by_uid('<uid>', adapter)
Group memberships
Memberships not granted
Sometimes you may think a particular user should be added to a GitLab group via LDAP group sync, but for some reason it’s not happening. You can check several things to debug the situation.
- Ensure LDAP configuration has a
group_base
specified. This configuration is required for group sync to work properly. - Ensure the correct LDAP group link is added to the GitLab group.
- Check that the user has an LDAP identity:
- Sign in to GitLab as an administrator user.
- On the top bar, select Main menu > Admin.
- On the left sidebar, select Overview > Users.
- Search for the user.
- Open the user by selecting their name. Do not select Edit.
- Select the Identities tab. There should be an LDAP identity with
an LDAP DN as the
Identifier
. If not, this user hasn’t signed in with LDAP yet and must do so first.
- You’ve waited an hour or the configured interval for the group to sync. To speed up the process, either go to the GitLab group Group information > Members and press Sync now (sync one group) or run the group sync Rake task (sync all groups).
If all of the above looks good, jump in to a little more advanced debugging in the rails console.
- Enter the rails console.
- Choose a GitLab group to test with. This group should have an LDAP group link already configured.
- Enable debug logging, find the above GitLab group, and sync it with LDAP.
- Look through the output of the sync. See example log output for how to read the output.
- If you still aren’t able to see why the user isn’t being added, query the LDAP group directly to see what members are listed.
- Is the user’s DN or UID in one of the lists from the above output? One of the DNs or UIDs here should match the ‘Identifier’ from the LDAP identity checked earlier. If it doesn’t, the user does not appear to be in the LDAP group.
Administrator privileges not granted
When Administrator sync has been configured but the configured users aren’t granted the correct administrator privileges, confirm the following are true:
- A
group_base
is also configured. - The configured
admin_group
in thegitlab.rb
is a CN, rather than a DN or an array. - This CN falls under the scope of the configured
group_base
. - The members of the
admin_group
have already signed into GitLab with their LDAP credentials. GitLab only grants administrator access to the users whose accounts are already connected to LDAP.
If all the above are true and the users are still not getting access,
run a manual group sync in the rails console and
look through the output to see what happens when
GitLab syncs the admin_group
.
Sync now button stuck in the UI
The Sync now button on the Group > Members page of a group can become stuck. The button becomes stuck after it is pressed and the page is reloaded. The button then cannot be selected again.
The Sync now button can become stuck for many reasons and requires debugging for specific cases. The following are two possible causes and possible solutions to the problem.
Invalid memberships
The Sync now button becomes stuck if some of the group’s members or requesting members are invalid. You can track progress on improving the visibility of this problem in a relevant issue. You can use a Rails console to confirm if this problem is causing the Sync now button to be stuck:
# Find the group in question
group = Group.find_by(name: 'my_gitlab_group')
# Look for errors on the Group itself
group.valid?
group.errors.map(&:full_messages)
# Look for errors among the group's members and requesters
group.requesters.map(&:valid?)
group.requesters.map(&:errors).map(&:full_messages)
group.members.map(&:valid?)
group.members.map(&:errors).map(&:full_messages)
A displayed error can identify the problem and point to a solution. For example, the Support Team has seen the following error:
irb(main):018:0> group.members.map(&:errors).map(&:full_messages)
=> [["The member's email address is not allowed for this group. Go to the group’s 'Settings > General' page, and check 'Restrict membership by email domain'."]]
This error showed that an Administrator chose to restrict group membership by email domain, but there was a typo in the domain. After the domain setting was fixed, the Sync now button functioned again.
Missing LDAP configuration on Sidekiq nodes
The Sync now button becomes stuck when GitLab is scaled over multiple nodes and the LDAP configuration is missing from
the /etc/gitlab/gitlab.rb
on the nodes running Sidekiq.
In this case, the Sidekiq jobs seem to disappear.
LDAP is required on the Sidekiq nodes because LDAP has multiple jobs that are run asynchronously that require a local LDAP configuration:
You can test whether missing LDAP configuration is the problem by running the Rake task to check LDAP on each node that is running Sidekiq. If LDAP is set up correctly on this node, it connects to the LDAP server and returns users.
To solve this issue, configure LDAP on the Sidekiq nodes. When configured, run the Rake task to check LDAP to confirm that the GitLab node can connect to LDAP.
Sync all groups
The output from a manual group sync can show you what happens when GitLab syncs its LDAP group memberships against LDAP.
Rails.logger.level = Logger::DEBUG
LdapAllGroupsSyncWorker.new.perform
Next, learn how to read the output.
Example console output after a group sync
Like the output from the user sync, the output from the manual group sync is also very verbose. However, it contains lots of helpful information.
Indicates the point where syncing actually begins:
Started syncing 'ldapmain' provider for 'my_group' group
The following entry shows an array of all user DNs GitLab sees in the LDAP server. These DNs are the users for a single LDAP group, not a GitLab group. If you have multiple LDAP groups linked to this GitLab group, you see multiple log entries like this - one for each LDAP group. If you don’t see an LDAP user DN in this log entry, LDAP is not returning the user when we do the lookup. Verify the user is actually in the LDAP group.
Members in 'ldap_group_1' LDAP group: ["uid=john0,ou=people,dc=example,dc=com",
"uid=mary0,ou=people,dc=example,dc=com", "uid=john1,ou=people,dc=example,dc=com",
"uid=mary1,ou=people,dc=example,dc=com", "uid=john2,ou=people,dc=example,dc=com",
"uid=mary2,ou=people,dc=example,dc=com", "uid=john3,ou=people,dc=example,dc=com",
"uid=mary3,ou=people,dc=example,dc=com", "uid=john4,ou=people,dc=example,dc=com",
"uid=mary4,ou=people,dc=example,dc=com"]
Shortly after each of the above entries, you see a hash of resolved member access levels. This hash represents all user DNs GitLab thinks should have access to this group, and at which access level (role). This hash is additive, and more DNs may be added, or existing entries modified, based on additional LDAP group lookups. The very last occurrence of this entry should indicate exactly which users GitLab believes should be added to the group.
Guest
, 20 is Reporter
, 30 is Developer
, 40 is Maintainer
and 50 is Owner
.Resolved 'my_group' group member access: {"uid=john0,ou=people,dc=example,dc=com"=>30,
"uid=mary0,ou=people,dc=example,dc=com"=>30, "uid=john1,ou=people,dc=example,dc=com"=>30,
"uid=mary1,ou=people,dc=example,dc=com"=>30, "uid=john2,ou=people,dc=example,dc=com"=>30,
"uid=mary2,ou=people,dc=example,dc=com"=>30, "uid=john3,ou=people,dc=example,dc=com"=>30,
"uid=mary3,ou=people,dc=example,dc=com"=>30, "uid=john4,ou=people,dc=example,dc=com"=>30,
"uid=mary4,ou=people,dc=example,dc=com"=>30}
It’s not uncommon to see warnings like the following. These indicate that GitLab would have added the user to a group, but the user could not be found in GitLab. Usually this is not a cause for concern.
If you think a particular user should already exist in GitLab, but you’re seeing this entry, it could be due to a mismatched DN stored in GitLab. See User DN and email have changed to update the user’s LDAP identity.
User with DN `uid=john0,ou=people,dc=example,dc=com` should have access
to 'my_group' group but there is no user in GitLab with that
identity. Membership will be updated when the user signs in for
the first time.
Finally, the following entry says syncing has finished for this group:
Finished syncing all providers for 'my_group' group
When all the configured group links have been synchronized, GitLab looks for any Administrators or External users to sync:
Syncing admin users for 'ldapmain' provider
The output looks similar to what happens with a single group, and then this line indicates the sync is finished:
Finished syncing admin users for 'ldapmain' provider
If administrator sync is not configured, you see a message stating as such:
No `admin_group` configured for 'ldapmain' provider. Skipping
Sync one group
Syncing all groups can produce a lot of noise in the output, which can be distracting when you’re only interested in troubleshooting the memberships of a single GitLab group. In that case, here’s how you can just sync this group and see its debug output:
Rails.logger.level = Logger::DEBUG
# Find the GitLab group.
# If the output is `nil`, the group could not be found.
# If a bunch of group attributes are in the output, your group was found successfully.
group = Group.find_by(name: 'my_gitlab_group')
# Sync this group against LDAP
EE::Gitlab::Auth::Ldap::Sync::Group.execute_all_providers(group)
The output is similar to that you get from syncing all groups.
Query a group in LDAP
When you’d like to confirm that GitLab can read a LDAP group and see all its members, you can run the following:
# Find the adapter and the group itself
adapter = Gitlab::Auth::Ldap::Adapter.new('ldapmain') # If `main` is the LDAP provider
ldap_group = EE::Gitlab::Auth::Ldap::Group.find_by_cn('group_cn_here', adapter)
# Find the members of the LDAP group
ldap_group.member_dns
ldap_group.member_uids
LDAP synchronization does not remove group creator from group
LDAP synchronization should remove an LDAP group’s creator from that group, if that user does not exist in the group. If running LDAP synchronization does not do this:
- Add the user to the LDAP group.
- Wait until LDAP group synchronization has finished running.
- Remove the user from the LDAP group.
User DN and email have changed
If both the primary email and the DN change in LDAP, GitLab cannot identify the correct LDAP record of a user. As a result, GitLab blocks that user. So that GitLab can find the LDAP record, update the user’s existing GitLab profile with at least either:
- The new primary email.
- DN values.
The following script updates the emails for all provided users so they aren’t blocked or unable to access their accounts.
Go to the rails console and then run:
# Each entry must include the old username and the new email
emails = {
'ORIGINAL_USERNAME' => 'NEW_EMAIL_ADDRESS',
...
}
emails.each do |username, email|
user = User.find_by_username(username)
user.email = email
user.skip_reconfirmation!
user.save!
end
You can then run a UserSync to sync the latest DN for each of these users.
Could not authenticate you from Ldapmain because "Unknown provider"
You can receive the following error when authenticating with an LDAP server:
Could not authenticate you from Ldapmain because "Unknown provider (ldapsecondary). available providers: ["ldapmain"]".
This error is caused when using an account that previously authenticated with an LDAP server that is renamed or removed from your GitLab configuration. For example:
- Initially,
main
andsecondary
are set inldap_servers
in GitLab configuration. - The
secondary
setting is removed or renamed tomain
. - A user attempting to sign in has an
identify
record forsecondary
, but that is no longer configured.
Use the Rails console to list affected users and check what LDAP servers they have identities for:
ldap_identities = Identity.where(provider: "ldapsecondary")
ldap_identities.each do |identity|
u=User.find_by_id(identity.user_id)
ui=Identity.where(user_id: identity.user_id)
puts "user: #{u.username}\n #{u.email}\n last activity: #{u.last_activity_on}\n #{identity.provider} ID: #{identity.id} external: #{identity.extern_uid}"
puts " all identities:"
ui.each do |alli|
puts " - #{alli.provider} ID: #{alli.id} external: #{alli.extern_uid}"
end
end;nil
You can solve this error in two ways.
Rename references to the LDAP server
This solution is suitable when the LDAP servers are replicas of each other, and the affected users should be able to sign in using a configured LDAP server. For example, if a load balancer is now used to manage LDAP high availability and a separate secondary sign-in option is no longer needed.
To rename references to the LDAP server that is no longer configured, run:
sudo gitlab-rake gitlab:ldap:rename_provider[ldapsecondary,ldapmain]
Remove the identity
records that relate to the removed LDAP server
With this solution, affected users can sign in with the configured LDAP servers and a new identity
record is created by GitLab. In a
Rails console, delete the ldapsecondary
identities:
ldap_identities = Identity.where(provider: "ldapsecondary")
ldap_identities.each do |identity|
puts "Destroying identity: #{identity.id} #{identity.provider}: #{identity.extern_uid}"
identity.destroy!
rescue => e
puts 'Error generated when destroying identity:\n ' + e.to_s
end; nil
Expired license causes errors with multiple LDAP servers
Using multiple LDAP servers requires a valid license. An expired license can cause:
-
502
errors in the web interface. -
The following error in logs (the actual strategy name depends on the name configured in
/etc/gitlab/gitlab.rb
):Could not find a strategy with name `Ldapsecondary'. Please ensure it is required or explicitly set it using the :strategy_class option. (Devise::OmniAuth::StrategyNotFound)
To resolve this error, you must apply a new license to the GitLab instance without the web interface:
- Remove or comment out the GitLab configuration lines for all non-primary LDAP servers.
- Reconfigure GitLab so that it temporarily uses only one LDAP server.
- Enter the Rails console and add the license key.
- Re-enable the additional LDAP servers in the GitLab configuration and reconfigure GitLab again.
Debugging Tools
LDAP check
The Rake task to check LDAP is a valuable tool to help determine whether GitLab can successfully establish a connection to LDAP and can get so far as to even read users.
If a connection can’t be established, it is likely either because of a problem with your configuration or a firewall blocking the connection.
- Ensure you don’t have a firewall blocking the connection, and that the LDAP server is accessible to the GitLab host.
- Look for an error message in the Rake check output, which may lead to your LDAP configuration to
confirm that the configuration values (specifically
host
,port
,bind_dn
, andpassword
) are correct. - Look for errors in the logs to further debug connection failures.
If GitLab can successfully connect to LDAP but doesn’t return any users, see what to do when no users are found.
GitLab logs
If a user account is blocked or unblocked due to the LDAP configuration, a
message is logged to application.log
.
If there is an unexpected error during an LDAP lookup (configuration error,
timeout), the sign-in is rejected and a message is logged to production.log
.
ldapsearch
ldapsearch
is a utility that allows you to query your LDAP server. You can
use it to test your LDAP settings and ensure that the settings you’re using
get you the results you expect.
When using ldapsearch
, be sure to use the same settings you’ve already
specified in your gitlab.rb
configuration so you can confirm what happens
when those exact settings are used.
Running this command on the GitLab host also helps confirm that there’s no obstruction between the GitLab host and LDAP.
For example, consider the following GitLab configuration:
gitlab_rails['ldap_servers'] = YAML.load <<-'EOS' # remember to close this block with 'EOS' below
main: # 'main' is the GitLab 'provider ID' of this LDAP server
label: 'LDAP'
host: '127.0.0.1'
port: 389
uid: 'uid'
encryption: 'plain'
bind_dn: 'cn=admin,dc=ldap-testing,dc=example,dc=com'
password: 'Password1'
active_directory: true
allow_username_or_email_login: false
block_auto_created_users: false
base: 'dc=ldap-testing,dc=example,dc=com'
user_filter: ''
attributes:
username: ['uid', 'userid', 'sAMAccountName']
email: ['mail', 'email', 'userPrincipalName']
name: 'cn'
first_name: 'givenName'
last_name: 'sn'
group_base: 'ou=groups,dc=ldap-testing,dc=example,dc=com'
admin_group: 'gitlab_admin'
EOS
You would run the following ldapsearch
to find the bind_dn
user:
ldapsearch -D "cn=admin,dc=ldap-testing,dc=example,dc=com" \
-w Password1 \
-p 389 \
-h 127.0.0.1 \
-b "dc=ldap-testing,dc=example,dc=com"
The bind_dn
, password
, port
, host
, and base
are all
identical to what’s configured in the gitlab.rb
.
Use ldapsearch with start_tls
encryption
The previous example performs an LDAP test in plaintext to port 389. If you are using start_tls
encryption, in
the ldapsearch
command include:
- The
-Z
flag. - The FQDN of the LDAP server.
You must include these because, during TLS negotiation, the FQDN of the LDAP server is evaluated against its certificate:
ldapsearch -D "cn=admin,dc=ldap-testing,dc=example,dc=com" \
-w Password1 \
-p 389 \
-h "testing.ldap.com" \
-b "dc=ldap-testing,dc=example,dc=com" -Z
Use ldapsearch with simple_tls
encryption
If you are using simple_tls
encryption (usually on port 636), include the following in the ldapsearch
command:
- The LDAP server FQDN with the
-H
flag and the port. - The full constructed URI.
ldapsearch -D "cn=admin,dc=ldap-testing,dc=example,dc=com" \
-w Password1 \
-H "ldaps://testing.ldap.com:636" \
-b "dc=ldap-testing,dc=example,dc=com"
For more information, see the official ldapsearch
documentation.
Using AdFind (Windows)
You can use the AdFind
utility (on Windows based systems) to test that your LDAP server is accessible and authentication is working correctly. AdFind is a freeware utility built by Joe Richards.
Return all objects
You can use the filter objectclass=*
to return all directory objects.
adfind -h ad.example.org:636 -ssl -u "CN=GitLabSRV,CN=Users,DC=GitLab,DC=org" -up Password1 -b "OU=GitLab INT,DC=GitLab,DC=org" -f (objectClass=*)
Return single object using filter
You can also retrieve a single object by specifying the object name or full DN. In this example we specify the object name only CN=Leroy Fox
.
adfind -h ad.example.org:636 -ssl -u "CN=GitLabSRV,CN=Users,DC=GitLab,DC=org" -up Password1 -b "OU=GitLab INT,DC=GitLab,DC=org" -f "(&(objectcategory=person)(CN=Leroy Fox))"
Rails console
The rails console is a valuable tool to help debug LDAP problems. It allows you to directly interact with the application by running commands and seeing how GitLab responds to them.
For instructions about how to use the rails console, refer to this guide.
Enable debug output
This provides debug output that shows what GitLab is doing and with what. This value is not persisted, and is only enabled for this session in the Rails console.
To enable debug output in the rails console, enter the rails console and run:
Rails.logger.level = Logger::DEBUG
Get all error messages associated with groups, subgroups, members, and requesters
Collect error messages associated with groups, subgroups, members, and requesters. This captures error messages that may not appear in the Web interface. This can be especially helpful for troubleshooting issues with LDAP group sync and unexpected behavior with users and their membership in groups and subgroups.
# Find the group and subgroup
group = Group.find_by_full_path("parent_group")
subgroup = Group.find_by_full_path("parent_group/child_group")
# Group and subgroup errors
group.valid?
group.errors.map(&:full_messages)
subgroup.valid?
subgroup.errors.map(&:full_messages)
# Group and subgroup errors for the members AND requesters
group.requesters.map(&:valid?)
group.requesters.map(&:errors).map(&:full_messages)
group.members.map(&:valid?)
group.members.map(&:errors).map(&:full_messages)
group.members_and_requesters.map(&:errors).map(&:full_messages)
subgroup.requesters.map(&:valid?)
subgroup.requesters.map(&:errors).map(&:full_messages)
subgroup.members.map(&:valid?)
subgroup.members.map(&:errors).map(&:full_messages)
subgroup.members_and_requesters.map(&:errors).map(&:full_messages)