The TOTP plugin provides a login flow that implements a bare-bones form of support for TOTP OATH tokens. Any software or hardware token supporting the TOTP algorithm should work. It can be used to piggyback on the existing Password flow's login form view, with an additional field to collect the token code, or it can be run separately with a dedicated view, usually via the IdP's Multi-Factor Authentication feature. In either case, the identity of the subject used to resolve token seeds is generally based on an earlier authentication step.
There are some limitations to understand about this add-on:
There is no support for enrollment of users and tokens or for managing those enrollments.
The default implementation provides a couple of options for looking up token seeds for subjects, a static/dummy option for testing and making use of the IdP's Attribute Resolver service to look up (and optionally decrypt) token information at runtime. An extension point does exist for providing alternative token sources, but this interface is not at this time a public API so should be viewed as unstable.
Taken together, this plugin is probably best viewed as either a feature for local extension or as a tool for some limited kinds of deployments where tokens can be managed by hand in some way much as SSH keys often are by small teams.
Installing this plugin will automatically enable or update the "idp.authn.TOTP" module since that is the only functionality provided. You are of course free to subsequently re-enable or disable the module by hand.
Token Creation and Testing
While this plugin lacks real management functionality, there is a simple command-line tool installed into bin for generating new token seeds and performing test validations.
The output includes the unencrypted, Base32-encoded seed, a URL that can be used to manually install the token into software authenticators, and a URL to a QR code generator for the token.
The same utility also includes a mode for validating codes by supplying both the seed and a code to check.
Installing/enabling this module will only populate a few files:
View template for dedicated prompt for token code
View template for reporting errors with code validation
Shell script for testing token code validation
Batch script for testing token code validation
Other than the user interface, which is fairly self-explanatory, the only configuration required is to define how to resolve token seeds to use while validating a particular subject's token codes. This is assumed to be the Attribute Resolver in most "real" cases, but it's also possible to define some static tokens for testing in a Spring configuration file (e.g., conf/global.xml).
After installation, a view (with some error handling examples) is provided to support a "two-stage" user interface where the token code is collected after an initial factor is completed that will supply the username to this flow. This is designed to work out of the box with a simple "Password flow, then TOTP flow" MFA configuration, much as the Duo flow works.
As an alternative, you may choose to "collapse" the collection process so that the Password login form includes the input field to collect the token code. This would perhaps be appropriate in cases when you impose TOTP as a second factor to all requests. The name of the input field is "tokencode" by default but is controllable with the idp.authn.TOTP.fieldName property. With this approach, the same underlying configuration is used but the flow will try and locate the form field before displaying the second view, and if for some reason the code fails, it will simply fall back to the second view to reprompt for the code.
One note about this: because of the way Spring messages work, unfortunately putting the input field on the Password collection view will not allow use of any Spring message identifiers that are pre-installed with this module, namely "idp.totp.field" for the field label. So you would need to add that to your own conf/messages.properties file or just inline the label.
Finally, for non-browser use, it's also possible to supply the code in a request header, "X-Shibboleth-TOTP" by default, also controllable with a property. Because of this, the flow is installed by default as supporting non-browser use, though if the header is missing there is no user interface to ask for one.
Source of Token Seeds
This plugin supports a public interface to obtain the token seeds for a specific user at runtime. There are two implementations provided, as described below.
The default behavior of the flow is to perform a special attribute resolution step to resolve an attribute named by the idp.authn.TOTP.tokenSeedAttribute property, defaulting to "tokenSeeds". Since the attribute name really shouldn't matter much, leaving it defaulted is usually the best choice but if you want to set the property you can put it in any of the existing property files, advisedly the conf/authn/authn.properties file added in V4.1.
The attribute itself must supply one or more string values containing the Base32-encoded token seed(s) to try and validate the supplied code against. Note these seeds cannot be encrypted at this stage. If you wish to store them in an encrypted form, it's possible to do so using the IdP using its "DataSealer" component that manages a list of keys shared amongst all the nodes of a cluster. Assuming this is done (details below), the DecryptedAttributeDefinition can be used to decrypt them on the fly to produce the resulting attribute.
DataSealer Definition and Use
While most IdPs include a DataSealer configured internally, this is typically not a good choice for this use case because the original use cases tend to assume the keys are being rotated frequently. To use a longer-term key, a dedicated component will typically be required. In order to facilitate using it to encrypt data using the command line, it's a good idea to define it in a separate Spring resource and load that into the Attribute Resolver's set of configuration resources in conf/services.xml.
Note that you will need to use the seckeygen utility yourself out of band to initialize the keystore and a key to use:
To use this from the command line to encrypt (and for example purposes, decrypt) data, assuming the file were named conf/totpsealer.xml:
DataSealer Command Example
$ cd /opt/shibboleth-idp
$ bin/sealer.sh --quiet conf/totpsealer.xml dec `bin/sealer.sh --quiet conf/totpsealer.xml enc data`
Note that the data produced by encryption is generally (though not absolutely required to be) Base64-encoded. This is not the same as how the token seed itself may be encoded, as the seed is in cleartext. The typical encoding for an encrypted seed would be a Base32-encoded seed, encrypted and encoded with Base64.
Primarily for testing, it's possible to supply a Spring bean containing a map of statically defined token seeds for users. The bean MUST be named shibboleth.authn.TOTP.SeedSource and will inherit from a parent bean named shibboleth.authn.TOTP.StaticSeedSource.
As an example, something like this in conf/global.xml will work:
In older versions and upgraded systems, this list is defined in conf/authn/general-authn.xml. In V4.1+, no default version of the list is provided and it may simply be placed in conf/global.xml if needed.