Skip to main content

Configure Single Sign-On

Teleport users can log in to servers, Kubernetes clusters, databases, web applications, and Windows desktops through their organization's Single Sign-On (SSO) provider.

  • Azure Active Directory (AD): Configure Azure Active Directory SSO for SSH, Kubernetes, databases, desktops and web apps.
  • Active Directory (ADFS): Configure Windows Active Directory SSO for SSH, Kubernetes, databases, desktops and web apps.
  • Google Workspace: Configure Google Workspace SSO for SSH, Kubernetes, databases, desktops and web apps.
  • GitHub: Configure GitHub SSO for SSH, Kubernetes, databases, desktops, and web apps.
  • GitLab: Configure GitLab SSO for SSH, Kubernetes, databases, desktops and web apps.
  • OneLogin: Configure OneLogin SSO for SSH, Kubernetes, databases, desktops and web apps.
  • OIDC: Configure OIDC SSO for SSH, Kubernetes, databases, desktops and web apps.
  • Okta: Configure Okta SSO for SSH, Kubernetes, databases, desktops and web apps.

How Teleport uses SSO

You can register your Teleport cluster as an application with your SSO provider. When a user signs in to Teleport, your SSO provider will execute its own authentication flow, then send an HTTP request to your Teleport cluster to indicate that authentication has completed.

Teleport authenticates users to your infrastructure by issuing short-lived certificates. After a user completes an SSO authentication flow, Teleport issues a short-lived certificate to the user. Teleport also creates a temporary user on the Auth Service backend.

Temporary user resources

After a user completes an SSO authentication flow, Teleport creates a temporary user resource for the user.

When a user signs in to Teleport with tsh login, they can configure the TTL of the user Teleport creates. Teleport enforces a limit of 30 hours (the default is 12 hours).

In the Teleport audit log, you will see an event of type user.create with information about the temporary user.

How can I inspect a temporary user resource?

You can inspect a temporary user resource created via your SSO integration by using the tctl command:

# Log in to your cluster with tsh so you can use tctl remotely
$ tsh login --proxy=example.teleport.sh
$ tctl get users/<username>

Here is an example of a temporary user resource created when the GitHub user myuser signed in to GitHub to authenticate to Teleport. This resource expires 12 hours after creation. The created_by field indicates that the resource was created by Teleport's GitHub SSO integration:

kind: user
metadata:
expires: "2022-06-15T04:02:34.586688054Z"
id: 0000000000000000000
name: myuser
spec:
created_by:
connector:
id: github
identity: myuser
type: github
time: "2022-06-14T16:02:34.586688441Z"
user:
name: system
expires: "0001-01-01T00:00:00Z"
github_identities:
- connector_id: github
username: myuser
roles:
- editor
- access
- auditor
status:
is_locked: false
lock_expires: "0001-01-01T00:00:00Z"
locked_time: "0001-01-01T00:00:00Z"
traits:
github_teams:
- my-team
kubernetes_groups: null
kubernetes_users: null
logins:
- root
version: v2

Certificates for SSO users

Along with creating a temporary user, Teleport issues SSH and X.509 certificates to a successfully authenticated SSO user's machine. This enables SSO users to authenticate to your cluster without Teleport needing to create a permanent record of them.

In the X.509 certificate, for example, the Subject field contains the same information defined in the temporary user resource. This enables Teleport to enforce RBAC rules for the authenticated user when they access resources in your cluster.

This is a Subject field for a certificate that Teleport issued for the GitHub user myuser, who signed in to a Teleport cluster via the GitHub SSO integration:

Subject: L=myuser/street=teleport.example.com/postalCode={"github_teams":["my-team"],"kubernetes_groups":null,"kubernetes_users":null,"logins":["root"]}, O=access, O=editor, O=auditor, CN=myuser/1.3.9999.1.7=teleport.example.com

The user belongs to the GitHub team my-team, which this Teleport cluster maps to the access, editor, and auditor roles in Teleport. (Read the guide for your SSO provider to determine how to configure role mapping.)

Inspecting your certificate subject

To inspect the contents of an X.509 certificate issued for your user after you sign in to Teleport via SSO, run the following commands:

$ TELEPORT_CLUSTER=<your cluster>
$ SSO_USER=<your username within your SSO provider>
$ openssl x509 -text -in ~/.tsh/keys/${TELEPORT_CLUSTER}/${SSO_USER}-x509.pem | grep "Subject:"

You can inspect an SSH certificate issued for your Teleport user with the following command:

$ ssh-keygen -L -f ~/.tsh/keys/${TELEPORT_CLUSTER}/${SSO_USER}-ssh/${TELEPORT_CLUSTER}-cert.pub

Multiple SSO providers

Since Teleport creates temporary users and issues short-lived certificates when a user authenticates via SSO, it is straightforward to integrate Teleport with multiple SSO providers. Besides the temporary user resource, no persistent backend data in Teleport is tied to a user's account with the SSO provider.

This also means that if one SSO provider becomes unavailable, the end user only needs to choose another SSO provider when signing in to Teleport. While the user may be locked out of their account with the first SSO provider, signing in via the second provider is sufficient for Teleport to issue a new certificate and grant the user access to your infrastructure.

Note that if the username of an SSO user already belongs to a user registered locally with the Auth Service (i.e., created via tctl users add), the SSO login will fail.

Logging in via SSO

Users can log in to Teleport via your SSO provider by executing a command similar to the following, using the --auth flag to specify the provider:

# This command will automatically open the default web browser and take a user
# through the login process with an SSO provider
$ tsh login --proxy=proxy.example.com --auth=github

The command opens a browser window and shows a URL the user can visit in the terminal to complete their SSO flow:

If browser window does not open automatically, open it by clicking on the link:
http://127.0.0.1:45235/055a310a-1099-43ea-8cf6-ffc41d88ad1f

Teleport will wait for up to 3 minutes for a user to authenticate. If authentication succeeds, Teleport will retrieve SSH and X.509 certificates and store them in the ~/.tsh/keys/<clustername> directory. The tool will also will add SSH cert to an SSH agent if there's one running.

Changing Callback Address

The callback address can be changed if calling back to a remote machine instead of the local machine is required:

# --bind-addr sets the host and port tsh will listen on, and --callback changes 
# what link is displayed to the user
$ tsh login --proxy=proxy.example.com --auth=github --bind-addr=localhost:1234 --callback https://remote.machine:1234

For this to work the hostname or CIDR of the remote machine that will be used for the callback will need to be allowed via your auth connector's client_redirect_settings:

kind: oidc
metadata:
name: example-connector
spec:
client_redirect_settings:
# a list of hostnames allowed for HTTPS client redirect URLs
# can be a regex pattern
allowed_https_hostnames:
- remote.machine
- '*.app.github.dev'
- '^\d+-[a-zA-Z0-9]+\.foo.internal$'
# a list of CIDRs allowed for HTTP or HTTPS client redirect URLs
insecure_allowed_cidr_ranges:
- '192.168.1.0/24'
- '2001:db8::/96'

Configuring SSO

Teleport works with SSO providers by relying on the concept of an authentication connector. An authentication connector is a configuration resource that controls how SSO users log in to Teleport—and which Teleport roles they will assume once they do.

This means that you can apply fine-grained RBAC policies to your Teleport cluster without needing to change the solution you use for on– and offboarding users.

Supported connectors

The following authentication connectors are supported:

TypeDescription
NoneIf no authentication connector is created, Teleport will use local authentication based user information stored
in the Auth Service backend. You can manage user data via the web UI Users page and the tctl users command.
samlThe SAML connector type uses the SAML protocol to authenticate users and query their group membership.
oidcThe OIDC connector type uses the OpenID Connect protocol to authenticate users
and query their group membership.
githubThe GitHub connector uses GitHub SSO to authenticate users and query their group membership.

Creating an authentication connector

Before you can create an authentication connector, you must enable authentication via that connector's protocol.

To set the default authentication type as saml, oidc, or github, create a cluster_auth_preference resource.

Create a file called cap.yaml:

kind: cluster_auth_preference
metadata:
name: cluster-auth-preference
spec:
# Set as saml, oidc, or github
type: saml|oidc|github
version: v2

Create the resource:

# Log in to your cluster with tsh so you can run tctl commands.
$ tsh login --proxy=mytenant.teleport.sh --user=myuser
$ tctl create -f cap.yaml

Next, define an authentication connector. Create a file called connector.yaml based on one of the following examples. Teleport Community Edition only supports GitHub as an SSO option.

#
# Example resource for a SAML connector
# This connector can be used for SAML endpoints like Okta
#
kind: saml
version: v2
metadata:
# the name of the connector
name: okta
spec:
# connector display name that will be appended to the title of "Login with"
# button on the cluster login screen so it will say "Login with Okta"
display: Okta
# SAML provider will make a callback to this URL after successful authentication
# cluster-url is the address the cluster UI is reachable at.
# The last segment of the URL must be identical to the connector metadata name.
acs: https://<cluster-url>/v1/webapi/saml/acs/new_saml_connector
# Controls whether IdP-initiated SSO is allowed. If false, all such requests will be rejected with an error.
allow_idp_initiated: false
attributes_to_roles:
- name: groups
value: okta-admin
roles:
- editor
- name: groups
value: okta-dev
roles:
- access
# Provides a path to the IdP metadata.
entity_descriptor_url: https://example.okta.com/app/your-app-id/sso/saml/metadata
# Optional SAML Single Logout endpoint. If set, logging out of Teleport
# will also log the user out of the SAML provider session.
single_logout_url: https://example.okta.com/app/your-app-id/slo/saml
client_redirect_settings:
# a list of hostnames allowed for HTTPS client redirect URLs
# can be a regex pattern
allowed_https_hostnames:
- remote.machine
- '*.app.github.dev'
- '^\d+-[a-zA-Z0-9]+\.foo.internal$'
# a list of CIDRs allowed for HTTP or HTTPS client redirect URLs
insecure_allowed_cidr_ranges:
- '192.168.1.0/24'
- '2001:db8::/96'

IdP-initiated SSO

Enabling the spec.allow_idp_initiated flag in SAML connectors allows users to log in to Teleport with one click from the dashboard provided by the IdP.

This feature is potentially unsafe and should be used with caution.

Enabling IdP-initiated login comes with notable security risks such as:

  • Possibility of replay attacks on the SAML payload giving an attacker a secret web session
  • Increased risk of session hijacking and impersonation attacks based on intercepting SAML communications
SAML Single Logout

Setting the spec.single_logout_url endpoint in SAML connectors enables SAML SLO (Single Logout). If enabled, upon logging out of Teleport, users will also be logged out of the SAML provider session, which may also log them out of any other non-Teleport applications which they are currently logged into using the same SAML provider.

For optimal user experience, we recommend keeping this disabled unless necessary.

Refer to your SAML provider's documentation for instructions on where to obtain this URL.

You may use entity_descriptor_url in lieu of entity_descriptor to fetch the entity descriptor from your IDP.

We recommend "pinning" the entity descriptor by including the XML rather than fetching from a URL.

Create the connector:

$ tctl create -f connector.yaml

User logins

Often it is required to restrict SSO users to their unique UNIX logins when they connect to Teleport Nodes. To support this:

  • Use the SSO provider to create a field called unix_login (you can use another name).
  • Make sure the unix_login field is exposed as a claim via SAML/OIDC.
  • Update a Teleport role to include the {{external.unix_login}} variable in the list of allowed logins:
kind: role
version: v5
metadata:
name: sso_user
spec:
allow:
logins:
- '{{external.unix_login}}'
node_labels:
'*': '*'

Provider-Specific Workarounds

Certain SSO providers may require or benefit from changes to Teleport's SSO flow. These provider-specific changes can be enabled by setting the spec.provider property of the connector definition to one of the following values to match your identity provider:

  • adfs (SAML): Required for compatibility with Active Directory (ADFS); refer to the full ADFS guide for details.
  • netiq (OIDC): Used to enable NetIQ-specific ACR value processing; refer to the OIDC guide for details.
  • ping (SAML and OIDC): Required for compatibility with Ping Identity (including PingOne and PingFederate).
  • okta (OIDC): Required when using Okta as an OIDC provider.

At this time, the spec.provider field should not be set for any other identity providers.

Working with an external email identity

Along with sending groups, an SSO provider will also provide a user's email address. In many organizations, the username that a person uses to log in to a system is the same as the first part of their email address, the "local" part.

For example, dave.smith@example.com might log in with the username dave.smith. Teleport provides an easy way to extract the first part of an email address so it can be used as a username. This is the {{email.local}} function.

If the email claim from the identity provider (which can be accessed via {{external.email}}) is sent and contains an email address, you can extract the "local" part of the email address before the @ sign like this: {{email.local(external.email)}}

Here's how this looks in a Teleport role:

kind: role
version: v5
metadata:
name: sso_user
spec:
allow:
logins:
# Extracts the local part of dave.smith@acme.com, so the login will
# now support dave.smith.
- '{{email.local(external.email)}}'
node_labels:
'*': '*'

Working with multiple SSO providers

Teleport can also support multiple connectors. For example, a Teleport administrator can define and create multiple connector resources using tctl create as shown above.

To see all configured connectors, execute this command on the Auth Server:

$ tctl get connectors

To delete/update connectors, use the usual tctl rm and tctl create commands as described in the Resources Reference.

If multiple authentication connectors exist, the clients must supply a connector name to tsh login via --auth argument:

# use "okta" SAML connector:
$ tsh --proxy=proxy.example.com login --auth=okta

# use local Teleport user DB:
$ tsh --proxy=proxy.example.com login --auth=local --user=admin

Refer to the following guides to configure authentication connectors of both SAML and OIDC types:

SSO customization

Use the display field in an authentication connector to control the appearance of SSO buttons in the Teleport Web UI.

ProviderYAMLExample
GitHubdisplay: GitHubgithub
Microsoftdisplay: Microsoftmicrosoft
Googledisplay: Googlegoogle
BitBucketdisplay: Bitbucketbitbucket
OpenIDdisplay: OktaOkta

Troubleshooting

Troubleshooting SSO configuration can be challenging. Usually a Teleport administrator must be able to:

  • Ensure that HTTP/TLS certificates are configured properly for both Teleport proxy and the SSO provider.
  • Be able to see what SAML/OIDC claims and values are getting exported and passed by the SSO provider to Teleport.
  • Be able to see how Teleport maps the received claims to role mappings as defined in the connector.

If something is not working, we recommend to:

  • Double-check the host names, tokens and TCP ports in a connector definition.

Using the Web UI

If you get "access denied" or other login errors, the number one place to check is the Audit Log. You can access it in the Activity tab of the Teleport Web UI.

Example of a user being denied because the role clusteradmin wasn't set up:

{
"code": "T1001W",
"error": "role clusteradmin is not found",
"event": "user.login",
"method": "oidc",
"success": false,
"time": "2019-06-15T19:38:07Z",
"uid": "cd9e45d0-b68c-43c3-87cf-73c4e0ec37e9"
}

Teleport does not show the expected Nodes

When Teleport's Auth Service receives a request to list Teleport Nodes (e.g., to display Nodes in the Web UI or via tsh ls), it only returns the Nodes that the current user is authorized to view.

For each Node in the user's Teleport cluster, the Auth Service applies the following checks in order and, if one check fails, hides the Node from the user:

  • None of the user's roles contain a deny rule that matches the Node's labels.
  • At least one of the user's roles contains an allow rule that matches the Node's labels.

If you are not seeing Nodes when expected, make sure that your user's roles include the appropriate allow and deny rules as documented in the Teleport Access Controls Reference.

When configuring SSO, ensure that the identity provider is populating each user's traits correctly. For a user to see a Node in Teleport, the result of populating a template variable in a role's allow.logins must match at least one of a user's traits.logins.

In this example a user will have usernames ubuntu, debian and usernames from the SSO trait logins for Nodes that have a env: dev label. If the SSO trait username is bob then the usernames would include ubuntu, debian, and bob.

kind: role
metadata:
name: example-role
spec:
allow:
logins: ['{{external.logins}}', ubuntu, debian]
node_labels:
'env': 'dev'
version: v5

Next steps

The roles we illustrated in this guide use external traits, which Teleport replaces with values from the single sign-on provider that the user used to authenticate with Teleport. For full details on how variable expansion works in Teleport roles, see the Teleport Access Controls Reference.