{project_name} supports logging in with an X.509 client certificate if you have configured the server to use mutual SSL authentication.
A typical workflow:
* A client sends an authentication request over SSL/TLS channel.
* During the SSL/TLS handshake, the server and the client exchange their x.509/v3 certificates.
* The container ({appserver_name}) validates the certificate PKIX path and the certificate expiration date.
* The x.509 client certificate authenticator validates the client certificate by using the following methods:
+
** Checks the certificate revocation status by using CRL or CRL Distribution Points.
** Checks the Certificate revocation status by using OCSP (Online Certificate Status Protocol).
** Validates whether the key in the certificate matches the expected key.
** Validates whether the extended key in the certificate matches the expected extended key.
+
* If any of the these checks fail, the x.509 authentication fails. Otherwise, the authenticator extracts the certificate identity and maps it to an existing user.
When the certificate maps to an existing user, the behavior diverges depending on the authentication flow:
* In the Browser Flow, the server prompts users to confirm their identity or sign in with a username and password.
* In the Direct Grant Flow, the server signs in the user.
IMPORTANT: Note that it is the responsibility of the web container to validate certificate PKIX path. X.509 authenticator on the
{project_name} side provides just the additional support for check the certificate expiration, certificate revocation status and key usage. If you are
using {project_name} deployed behind reverse proxy, make sure that your reverse proxy is configured to validate PKIX path. If you
do not use reverse proxy and users directly access the {appserver_name}, you should be fine as {appserver_name} makes sure that PKIX path is validated as long
{project_name} extracts the certificate identity from Subject DN or Issuer DN by using a regular expression as a filter. For example, this regular expression matches the email attribute:
The regular expression filtering applies if the `Identity Source` is set to either `Match SubjectDN using regular expression` or `Match IssuerDN using regular expression`.
The certificate identity mapping can map the extracted user identity to an existing user's username, email, or a custom attribute whose value matches the certificate identity. For example, setting `Identity source` to _Subject's email_ or `User mapping method` to _Username or email_ makes the X.509 client certificate authenticator use the email attribute in the certificate's Subject DN as the search criteria when searching for an existing user by username or by email.
* If you disable *Login with email* at realm settings, the same rules apply to certificate authentication. Users are unable to log in by using the email attribute.
* Using `Certificate Serial Number and IssuerDN` as an identity source requires two custom attributes for the serial number and the IssuerDN.
* `SHA-256 Certificate thumbprint` is the lowercase hexadecimal representation of SHA-256 certificate thumbprint.
* Using `Full certificate in PEM format` as an identity source is limited to the custom attributes mapped to external federation sources, such as LDAP. {project_name} cannot store certificates in its database due to length limitations, so in the case of LDAP, you must enable `Always Read Value From LDAP`.
Defines whether to use canonical format to determine a distinguished name. The official link:https://docs.oracle.com/javase/8/docs/api/javax/security/auth/x500/X500Principal.html#getName-java.lang.String-[Java API documentation] describes the format. This option affects the two User Identity Sources _Match SubjectDN using regular expression_ and _Match IssuerDN using regular expression_ only. Enable this option when you set up a new {project_name} instance. Disable this option to retain backward compatibility with existing {project_name} instances.
*Enable Serial Number hexadecimal representation*::
Represent the link:https://datatracker.ietf.org/doc/html/rfc5280#section-4.1.2.2[serial number] as hexadecimal. The serial number with the sign bit set to 1 must be left padded with 00 octet. For example, a serial number with decimal value _161_, or _a1_ in hexadecimal representation is encoded as _00a1_, according to RFC5280. See link:https://datatracker.ietf.org/doc/html/rfc5280#appendix-B[RFC5280, appendix-B] for more details.
A regular expression to use as a filter for extracting the certificate identity. The expression must contain a single group.
*User Mapping Method*::
Defines the method to match the certificate identity with an existing user. _Username or email_ searches for existing users by username or email. _Custom Attribute Mapper_ searches for existing users with a custom attribute that matches the certificate identity. The name of the custom attribute is configurable.
A custom attribute whose value matches against the certificate identity. Use multiple custom attributes when attribute mapping is related to multiple values, For example, 'Certificate Serial Number and IssuerDN'.
Check the revocation status of the certificate by using the Certificate Revocation List. The location of the list is defined in the *CRL file path* attribute.
By default the OCSP check must return a positive response in order to continue with a successful authentication. Sometimes however this check can be inconclusive: for example, the OCSP server could be unreachable, overloaded, or the client certificate may not contain an OCSP responder URI. When this setting is turned ON, authentication will be denied only if an explicit negative response is received by the OCSP responder and the certificate is definitely revoked. If a valid OCSP response is not available the authentication attempt will be accepted.
Verifies the certificate's KeyUsage extension bits are set. For example, "digitalSignature,KeyEncipherment" verifies if bits 0 and 2 in the KeyUsage extension are set.
Leave this parameter empty to disable the Key Usage validation. See link:https://datatracker.ietf.org/doc/html/rfc5280#section-4.2.1.3[RFC5280, Section-4.2.1.3] for more information. {project_name} raises an error when a key usage mismatch occurs.
Verifies one or more purposes defined in the Extended Key Usage extension. See link:https://datatracker.ietf.org/doc/html/rfc5280#section-4.2.1.12[RFC5280, Section-4.2.1.12] for more information. Leave this parameter empty to disable the Extended Key Usage validation. {project_name} raises an error when flagged as critical by the issuing CA and a key usage extension mismatch occurs.
Verifies one or more policy OIDs as defined in the Certificate Policy extension. See link:https://datatracker.ietf.org/doc/html/rfc5280#section-4.2.1.4[RFC5280, Section-4.2.1.4]. Leave the parameter empty to disable the Certificate Policy validation. Multiple policies should be separated using a comma.
*Certificate Policy Validation Mode*::
When more than one policy is specified in the `Validate Certificate Policy` setting, it decides whether the matching should check for all requested policies to be present, or one match is enough for a successful authentication. Default value is `All`, meaning that all requested policies should be present in the client certificate.
If enabled, X.509 client certificate authentication does not prompt the user to confirm the certificate identity. {project_name} signs in the user upon successful authentication.
If set, the client certificate trust chain will be always verified at the application level using the certificates present in the configured trust store. This can be useful if the underlying web server does not enforce client certificate chain validation, for example because it is behind a non-validating load balancer or reverse proxy, or when the number of allowed CAs is too large for the mutual SSL negotiation (most browsers cap the maximum SSL negotiation packet size at 32767 bytes, which corresponds to about 200 advertised CAs). By default this option is off.
