7.3. Authentication policies

The scope authentication gives you more detailed possibilities to authenticate the user or to define what happens during authentication.

Technically the authentication policies apply to the REST API Validate endpoints and are checked using Policy Module and Policy Decorators.

The following actions are available in the scope authentication:

7.3.1. otppin

type: string

This action defines how the fixed password part during authentication should be validated. Each token has its own OTP PIN, but you can choose how the authentication should be processed:

otppin=tokenpin

This is the default behaviour. The user needs to pass the OTP PIN concatenated with the OTP value.

otppin=userstore

The user needs to pass the user store password concatenated with the OTP value. It does not matter if the OTP PIN is set or not. If the user is located in an Active Directory the user needs to pass his domain password together with the OTP value.

Note

The domain password is checked with an LDAP bind right at the moment of authentication. So if the user is locked or the password was changed authentication will fail.

otppin=none

The user does not have to pass any fixed password. Authentication is only done via the OTP value.

7.3.2. passthru

type: str

If the user has no token assigned, he will be authenticated against the userstore or against the given RADIUS configuration. I.e. the user needs to provide the LDAP- or SQL-password or valid credentials for the RADIUS server.

Note

This is a good way to do a smooth enrollment. Users having a token enrolled will have to use the token, users not having a token, yet, will be able to authenticate with their domain password.

It is also a way to do smooth migrations from other OTP systems. The authentication request of users without a token is forwarded to the specified RADIUS server.

Note

The passthru policy overrides the authorization policy for tokentype. I.e. a user may authenticate due to the passthru policy (since he has no token) although a tokentype policy is active!

Warning

If the user has the right to delete his tokens in selfservice portal, the user could delete all his tokens and then authenticate with his static password again.

7.3.3. passthru_assign

type: str

This policy is only evaluated, if the policy passthru is set. If the user is authenticated against a RADIUS server, then privacyIDEA splits the sent password into PIN and OTP value and tries to find an unassigned token, that is in the user’s realm by using the OTP value. If it can identify this token, it assigns this token to the user and sets the sent PIN.

The policy is configured with a string value, that contains * the position of the PIN * the OTP length and * the number of OTP values tested for each unassigned token (optional, default=100).

Examples are

  • 8:pin would be an eight digit OTP value followed by the PIN

  • pin:6:10000 would be the PIN followed by an 6 digit OTP value, 10.000 otp values would be checked for each token.

Note

This method can be used to automatically migrated tokens from an old system to privacyIDEA. The administrator needs to import all seeds of the old tokens and put the tokens in the user’s realm.

Warning

This can be very time consuming if the OTP values to check is set to high!

7.3.4. passOnNoToken

type: bool

If the user has no token assigned an authentication request for this user will always be true.

Warning

Only use this if you know exactly what you are doing.

7.3.5. passOnNoUser

type: bool

If the user does not exist, the authentication request is successful.

Warning

Only use this if you know exactly what you are doing.

7.3.6. smstext

type: string

This is the text that is sent via SMS to the user trying to authenticate with an SMS token. You can use the tags <otp> and <serial>.

Starting with version 2.20 you can use the tag {challenge}. This will add the challenge data that was passed in the first authentication request in the challenge parameter. This could contain banking transaction data.

Default: <otp>

7.3.7. smsautosend

type: bool

A new OTP value will be sent via SMS if the user authenticated successfully with his SMS token. Thus the user does not have to trigger a new SMS when he wants to login again.

7.3.8. emailtext

type: string

This is the text that is sent via Email to be used with Email Token. This text should contain the OTP value.

The text can contain the following tags, that will be filled:

  • {serial} the serial number of the token.

  • {user} the given name of the token owner.

  • {givenname} the given name of the token owner.

  • {surname} the surname of the token owner.

  • {username} the loginname of the token owner.

  • {userrealm} the realm of the token owner.

  • {tokentype} the type of the token.

  • {recipient_givenname} the given name of the recipient.

  • {recipient_surname} the surname of the recipient.

  • {time} the current server time in the format HH:MM:SS.

  • {date} the current server date in the format YYYY-MM-DD

Starting with version 2.20 you can use the tag {challenge}. This will add the challenge data that was passed in the first authentication request in the challenge parameter. This could contain banking transaction data.

Default: <otp>

You can also provide the filename to an email template. The filename must be prefixed with file: like file:/etc/privacyidea/emailtemplate.html. The template is an HTML file.

Note

If a message text is supplied directly, the email is sent as plain text. If the email template is read from a file, a HTML-only email is sent instead.

7.3.9. emailsubject

type: string

This is the subject of the Email sent by the Email Token. You can use the same tags as mentioned in emailtext.

Default: Your OTP

7.3.10. emailautosend

type: bool

If set, a new OTP Email will be sent, when successfully authenticated with an Email Token.

7.3.11. mangle

type: string

The mangle policy can mangle the authentication request data before they are processed. I.e. the parameters user, pass and realm can be modified prior to authentication.

This is useful if either information needs to be stripped or added to such a parameter. To accomplish that, the mangle policy can do a regular expression search and replace using the keyword user, pass (password) and realm.

A valid action could look like this:

action: mangle=user/.*(.{4})/user\\1/

This would modify a username like “userwithalongname” to “username”, since it would use the last four characters of the given username (“name”) and prepend the fixed string “user”.

This way you can add, remove or modify the contents of the three parameters. For more information on the regular expressions see 1.

Note

You must escape the backslash as \ to refer to the found substrings.

Example: A policy to remove whitespace characters from the realm name would look like this:

action: mangle=realm/\\s//

Example: If you want to authenticate the user only by the OTP value, no matter what OTP PIN he enters, a policy might look like this:

action: mangle=pass/.*(.{6})/\\1/

Example: If you want to strip a string from the front of a username, for example to have “admin_username” resolve to just “username”, it would look like this:

action: mangle=user/admin_(.*)/\\1/

7.3.12. challenge_response

type: string

This is a list of token types for which challenge response can be used during authentication. The list is separated by whitespaces like “hotp totp”.

7.3.13. u2f_facets

type: string

This is a white space separated list of domain names, that are trusted to also use a U2F device that was registered with privacyIDEA.

You need to specify a list of FQDNs without the https scheme like:

“host1.example.com host2.exmaple.com firewall.example.com”

For more information on configuring U2F see U2F.

1

https://docs.python.org/2/library/re.html

7.3.14. reset_all_user_tokens

type: bool

If a user authenticates successfully all failcounter of all of his tokens will be reset. This can be important, if using empty PINs or otppin=None.

7.3.15. auth_cache

type: string

The Authentication Cache caches the credentials of a successful authentication and allows to use the same credentials - also with an OTP value - for the specified amount of time.

The time to cache the credentials can be specified like “4h”, “5m”, “2d” (hours, minutes days) or “4h/5m”. The notation 4h/5m means, that credentials are cached for 4 hours, but only may be used again, if every 5 minutes the authentication occurs. If the authentication with the same credentials would not occur within 5 minutes, the credentials can not be used anymore.

In future implementations the caching of the credentials could also be dependent on the clients IP address and the user agent.

Note

Cache entries are written to the database table authcache. Please note that expired entries are automatically deleted only when the user attempts to log in with the same expired credentials again. In all other cases, expired entries need to be deleted from this table manually by running:

pi-manage authcache cleanup --minutes MIN

which deletes all cache entries whose last authentication has occurred at least MIN minutes ago. As an example:

pi-manage authcache cleanup --minutes 300

will delete all authentication cache entries whose last authentication happened more than 5 hours ago.

It may make sense to create a cronjob that periodically cleans up old authentication cache entries.

Note

The AuthCache only works for user authentication, not for authentication with serials.

7.3.16. push_text_on_mobile

type: string

This is the text that should be displayed on the push notification during the login process with a Push Token. You can choose different texts for different users or IP addresses. This way you could customize push notifications for different applications.

7.3.17. push_title_on_mobile

type: string

This is the title of the push notification that is displayed on the user’s smartphone during the login process with a Push Token.

7.3.18. push_wait

type: int

This can be set to a number of seconds. If this is set, the authentication with a push token is only performed via one request to /validate/check. The HTTP request to /validate/check will wait up to this number of seconds and check, if the push challenge was confirmed by the user.

This way push tokens can be used with any non-push-capable applications.

Sensible numbers might be 10 or 20 seconds.

Note

This behaviour can interfere with other tokentypes. Even if the user also has a normal HOTP token, the /validate/check request will only return after this number of seconds.

Warning

Using simple webserver setups like Apache WSGI this actually can block all available worker threads, which will cause privacyIDEA to become unresponsive if the number of open PUSH challenges exceeds the number of available worker threads!

7.3.20. webauthn_allowed_transports

type: string

This action determines, which transports may be used to communicate with the authenticator, during authentication. For instance, if the authenticators used support both an USB connection and NFC wireless communication, they can be limited to USB only using this policy. The allowed transports are given as a space-separated list.

The default is to allow all transports (equivalent to a value of usb ble nfc internal lightning).

7.3.21. webauthn_timeout

type: integer

This action sets the time in seconds the user has to confirm an authentication request on his WebAuthn authenticator.

This is a client-side setting, that governs how long the client waits for the authenticator. It is independent of the time for which a challenge for a challenge response token is valid, which is governed by the server and controlled by a separate setting. This means, that if you want to increase this timeout beyond two minutes, you will have to also increase the challenge validity time, as documented in Challenge Validity Time.

This setting is a hint. It is interpreted by the client and may be adjusted by an arbitrary amount in either direction, or even ignored entirely.

The default timeout is 60 seconds.

Note

If you set this policy you may also want to set webauthn_timeout.

7.3.22. webauthn_user_verification_requirement

type: string

This action configures whether the user’s identity should be checked when authenticating with a WebAuthn token. If this is set to required, any user signing in with their WebAuthn token will have to provide some form of verification. This might be biometric identification, or knowledge-based, depending on the authenticator used.

This defaults to preferred, meaning user verification will be performed if supported by the token.

Note

User verification is different from user presence checking. The presence of a user will always be confirmed (by asking the user to take action on the token, which is usually done by tapping a button on the authenticator). User verification goes beyond this by ascertaining, that the user is indeed the same user each time (for example through biometric means), only set this to required, if you know for a fact, that you have authenticators, that actually support some form of user verification (these are still quite rare in practice).

Note

If you configure this, you will likely also want to configure webauthn_user_verification_requirement.