Updated 14SEPT2022 - I moved the GitHub link over to the official Jamf Github - https://github.com/jamf/jamfconnect/tree/main/azure_conditional_access will have the latest until the official Jamf Connect docs get updated.
Updated 14JUL2022 - The github link below has been updated with some steps removed for version 2.13 or greater of Jamf Connect, details about custom ROPG scopes in the menu bar, notes on how the login may still show a failure after doing this but that's fine. https://www.jamf.com/blog/how-to-azure-conditional-access-and-jamf-connect/ will supersede instructions currently on the Jamf Blog.
Updated 14JAN2022 - The github link below has been updated to simplify the setup of the application registrations in Azure and allows for full testing in Jamf Connect Configuration before deploying to a test machine.
https://www.jamf.com/blog/how-to-azure-conditional-access-and-jamf-connect/ - Updated instructions posted to Jamf Blog.
UPDATED: 10 December 2021 - Includes information on how to create a custom scope for Microsoft Azure Conditional Access policies.
openid profile email
The Open ID Connect 2.0 specification uses these default scopes to obtain an access or identity token for a client application. Consequently, in its default configuration, Jamf Connect login uses the "openid profile email" scope in its authentication requests, and the only way to apply a CA policy in this default behavior is to apply the policy to "All cloud apps" with NO exceptions applied or the CA policy will break.
AADSTS50076: Due to a configuration change made by your administrator, or because you moved to a new location, you must use multi-factor authentication to access [application UUID]
AADSTS50126: Error validating credentials due to invalid username or password.
Navigate to Azure Active Directory → Enterprise Applications and select the name of your Jamf Connect application in Azure. Navigate to Activity → Sign-ins to open user usage logs.
Shown above are two logins which appear to be failures. Under the “Authentication required” column, the first login says “Multi-factor authentication”. Clicking on the row will pull up additional details about the login attempt.
Under Authentication Details, the “Result detail” will let an administrator determine if the login was successful or a failure. In this example, the login was a success - the Result detail shows that the “User did not pass the MFA challenge (non interactive).” This login can be interpreted in that the user was required to use MFA by either a Conditional Access policy or through Azure Multi-factor authentication.
In the second example, a user with MFA required failed to enter their correct password:
The Authentication required column shows “Single-factor authentication” and the Authentication Details show “Invalid username or password or Invalid on-premise username or password.” While the user is required to use Multi-factor authentication, the user failed the first, single factor and thus was never prompted for MFA.
To eliminate the inaccurate "failed responses" for ROPG, admins must remove any "All cloud apps" scoped requirements for multi-factor authentication and create a custom scope for Jamf Connect login.
Follow the guide in Azure_Conditional_Access_and_Jamf_Connect.pdf (source: GitHub) for step by step instructions.
Several Conditional Access Grant policies can create unacceptable behavior to access a client device:
In the above cases, a user would be unable to log into a client machine to fix the issue of being out of compliance by running a Jamf Pro policy to get the device back into compliance.
In the above cases, the Jamf Connect software is not of the Microsoft apps accessing a specific service (for example, Microsoft Outlook accessing O365 mail), and all access would be blocked.
Administrators are recommended to carefully read conditional access polices and conditions applied to avoid locking users out of client devices inadvertently.
I've been advised by Jamf support not to implement this.
"We would advise against using the KB. Right now due to limitations in Azure, we can not exclude it from the CA policy - so those errors are expected. We can mitigate the volume of errors by changing the network check to something longer such as 30 minutes to an hour, and set "checkOnNetworkChange" to false."
I have followed these steps exactly and ensured everything has been set properly. I am able to successfully be prompted for MFA via the login window and my sign-in logs show the login with the OICD application to be successful, however the login shows the error message "An error occurred. Contact your it administrator". Not sure what I'm missing.
I've got an update I JUST posted to GitHub (the official jamf blog will be updated soon too) which simplifies the process, requires only two App registrations, and can be fully tested inside of Jamf Connect Configuration. Hit https://github.com/sean-rabbitt/jamf-connect-azure-conditional-access/blob/main/Azure_Conditional_Ac...with the updated deets.
Thank you so much for the updated document! Glad I caught you just in time :).
After making the changes, the OICD test worked in the configuration application, however ROPG did not. It appears to require the client secret. After entering my client secret in both the IdP and Connect tabs, it proceeds to error out for OICD and ROPG. I can gather additional troubleshooting info if necessary. Would like to clarify if I'm missing anything about the client secret first.
Client secret should not be required. If you defined a client secret in the public app (the one without any API permissions at all), you can either remove it or add that client secret to the OIDC and ROPG configuration in Jamf Connect Configuration to test it. I was not testing with a client secret.
Sample configuration profile with this setup:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<string>[YOUR PUBLIC APPLICATION UUID STRING HERE]</string>
<string>https://login.microsoftonline.com/[YOUR TENANT UUID STRING HERE]/v2.0/.well-known/openid-configuration</string>
<string>[YOUR PUBLIC APPLICATION UUID STRING HERE]</string>
<string>api://[YOUR RANDOMLY GENERATED API IDENTIFIER HERE]/jamfconnect+openid+profile+email</string>
<string>[YOUR TENANT UUID STRING HERE]</string>
<string>https://login.microsoftonline.com/[YOUR TENANT UUID STRING HERE]/.well-known/openid-configuration</string>
<string>[YOUR TENANT UUID STRING HERE]</string>
Hi there !
I've worked on many Connect implementations with Azure, and used your article as a reference many times.
Now, I was wondering, why would the ROPG part of JAMF Connect simply not ask for MFA when getting the AADSTS50076 answer ? People would not really be surprised to get a prompt to confirm MFA when changing location.
This would simplify a lot the discussions with organisations whhich really want to apply MFA to any app, and are annoyed by creating a second app which is MFA-exempt.
The ROPG check that happens every 60 minutes is supposed to be a "non-interactive" login. End users may become... unhappy if they're prompted to MFA every 60 minutes.
The real trick here is that the request for MFA response AADSTS50076 will ONLY come up if the non-interactive send of the user name and password is correct. At that point, Jamf Connect's menu bar agent knows the password is correct and we don't really need the identity/access token. We just want to know the password is still correct.
This is the purpose of the instructions above - allow there to be a app registration that is not subject to MFA requirements to prevent the "failed" login.
To me, an MFA request every 60 minutes risks a larger danger of training a user to accept every MFA request blindly rather than improve security.
When I attempt to test the ROPG login with the Jamf Connect Configuration application, it displays the error AADSTS900144: The request body must contain the following parameter: 'resource'. Fortunately, the OIDC test is working great with MFA. I have yet to test in an actual install.
I'm not sure if I'm missing a step during this setup or if I have something misconfigured. I'd appreciate any help with this. Thanks!
In regards to the following:
(H3) Step Four: Remove any Conditional Access policies applied to All cloud apps Navigate to
portal.azure.com → Azure Conditional Access. Examine any application applied to the scope of “All
cloud apps”. Either set “Enable policy” to “Oﬀ” for any application that has a Grant of “Require
multi-factor authentication” or modify the “Cloud apps or actions” to specifically list resources
that should have MFA applied.Applying a policy to require MFA for “All cloud apps” will cause the ROPG application in the next
step to inaccurately show failed logins in the Azure sign-in
Rather than modify "All cloud apps" to include all apps except Jamf, can Jamf just be added as an exclusion to existing CA's that require MFA? Would it just be "Jamf Connect - Conditional Access Policy API" that needs to be excluded?
So to be clear - this will result in the following?
Jamf Connect logging failures in Azure with the 50076 sign in error code, even when the MFA has prompted and been accepted by the Mac? Is there a way to have the 50076 errors in Azure show as successes instead of failures?
We're worried about the constant failures resulting in the triggering of a block out for the user.
We're on 2.14 and have configured things to spec per the instructions and the chat here. Just want to make sure I can properly communicate what's occurring back to the powers that be. Or if we need to tweak something - that we do so. Thanks in advance for any input.
The error isn't really an error - it's just an interrupted login. By the time the user needs to MFA in the password check to get an access token, Jamf Connect already knows the password is right.
If we think of the menu bar agent as the thing that checks the local client password against the cloud password, we don't really need the full access/ID/refresh tokens that would be sent down to the client if the user actually did the MFA. JC would look at that and say, "Thanks, but..." and then pitch those into the bit bucket anyway. JC got what it needed - was the password right? If you're prompting for MFA, then the password HAD to be right.
What the "complex but exacting" method of setting up two apps does will exclude the "public" endpoint, the one that does ROPG, from needing MFA at all. This will eliminate the 50076 errors. When the user is doing an interactive login (aka logging into the Mac, seeing the Azure login web page) then they'll be prompted for MFA because the scope requests access to a resource that needs MFA.
It's confusing, but it eliminates the errors in the logs.