Zoho Banner September 2011

OK, the title has a whole bunch of acronyms which may not be entirely familiar. Actually…if we’re being really picky I should probably say a whole bunch of initialisms, but that would digress into a whole different article when a perfectly good Wikipedia article already exists for that. :-)

Anyway, PTA is the accepted short form of Pass-Through Authentication – one of the range of authentication options available with Azure Active Directory.

AADJ stands for Azure Active Directory Domain Join(ed).  This is a state for a Windows 10 machine in which it is joined to Azure Active Directory for a given tenant organisation.  It is materially different to Azure AD Device Registration and Hybrid Azure AD Join, as neatly described here.

If you’re already set up with Azure AD Connect, have AADJ devices and are using PTA for your user sign-ins then you should be aware of an important limitation with respect to the “User must change password at next log on” flag.  The flag itself is set on the user object in on-premises Active Directory.  If you’ve been around a while, you’ll already be familiar with the setting - it looks like this:

1

 

The setting is used in a number of organisations to deal with the following situations:

  • User has forgotten their password and the Service Desk assigns them a temporary password to get them going again
  • New hires who are assigned a temporary password to start them off

So what’s the problem with setting this flag in our PTA, AADJ scenario?  Well, quite simply the user won’t be able to sign-in to their Windows 10 machine.  Instead of the user being prompted to change their password when entering the credentials that include the temporary password, the user sees the generic, “The user name or password is incorrect. Try again”.

Hopefully Microsoft will provide a resolution to this in the near future.  At the time of writing the behaviour is seen as “by design” in so far as the error generated on the DC to which the credentials have been passed cannot be successfully translated back to the point where the sign in attempt occurs.

 

Configurable token lifetimes for Azure Active Directory (AAD) have been available for while now, although the feature is still in public preview.  This article provides details of how to create an access token lifetime policy and how to apply it to an application federated with AAD using SAML 2.0.

Before we get started with this, we need to ensure you have the correct (i.e. Preview) version of the AAD Powershell Module.  The current link for this is:

https://www.powershellgallery.com/packages/AzureADPreview/2.0.0.17

Note that the module is subject to change, so search for the latest version.

The default Access Token Lifetime Policy that applies to SAML2 tokens is one hour as described in this article.

3

Ok, let’s go ahead and create a new Token Lifetime Policy.  To do this we are going to use the New-AzureADPolicy cmdlet, as shown in the example below.

New-AzureADPolicy-Definition @(‘{“TokenLifetimePolicy”:{“Version”:1,”AccessTokenLifetime”:”12:00:00″}}’) -DisplayName“12HourTLP”-IsOrganizationDefault$false-Type“TokenLifetimePolicy”

In this example, I have set the token lifetime to 12 hours.  Now this is just an example, you will need to consider the security implications of whatever policy you create.  Here’s the output.

1

We will need to make a note of the Id (GUID value) of the new policy as we will need this later.

The next step if to identify the service principal associated with your SAML-enabled application.  This uses the Get-AzureADServicePrincipal cmdlet, as follows:

Get-AzureADServicePrincipal-SearchString“My”

We can run the cmdlet without the searchstring parameter, but that tends to return a lot of results for us to pick our way through. Here’s what the output looks like.  Again, we should make a note of the ObjectID value as we will need this later.

2

Now you check which policies currently apply to your service principal.  We use the Get-AzureADServicePrincipalPolicy cmdlet to do this using the ObjectID of the service principal for our application.

Get-AzureADServicePrincipalPolicy-Id1911f64f-9d76-4ebf-9fcb-b3814e2e5e21

In this example, the output shows that a TokenIssuancePolicy is applied, but no TokenLifeTimePolicy – so we can assume that the default TokenLifeTimePolicy of 1 hour is in play.

4

Now we can go ahead and apply our newly created TokenLifeTimePolicy to the service principal representing our application.  This uses the Add-AzureADServicePrincipalPolicy cmdlet. The “Id” parameter needs to the ObjectID of our service principal, while the “RefObjectId” parameter needs the GUID of the Token Lifetime Policy we created earlier. And, yes, it can be confusing!

 

Add-AzureADServicePrincipalPolicy-Id1911f64f-9d76-4ebf-9fcb-b3814e2e5e21-RefObjectId74f4296d-fcdb-4c72-b434-b1628adef47b

Note that, if successful, this cmdlet returns no output.

5

We can now re-run the cmdlet to check which policies have been applied to our service principal.

6

As you can see, our 12HourTLP policy is now applied.

This is all very well, but how can we determine whether the policy is actually in effect or not?  One option is to sign-in to the application and wait for 12 hours to roll over.  If you have luxury of time for this then you clearly aren’t as busy as I am!  A better option is to examine the SAML Response XML using a SAML inspection tool such as the SAML Chrome Panel extension for the Chrome browser. Once you have the Response XML, look at the Conditions node and confirm that the NotBefore and NotOnOrAfter values show a 12 hour difference – see example below.

7

That’s it really.  In this article you have hopefully learned how to create a new Access Token Lifetime Policy as well as how to apply it to an existing SAML 2.0 application that is leveraging AAD as the Identity Provider (IdP).

Until next time!

Tony

This article describes how to configure Azure Active Directory as the SAML Identity Provider (IdP) to change the default AWS Console timeout from 1 hour to a different value.

It seems there has been a lot of discussion about how to change the timeout and there is no clear documentation from AWS how to achieve this with Azure AD.  As an example of the confusion, have a look at this discussion thread:

https://forums.aws.amazon.com/message.jspa?messageID=733264

Some good guidance is provided on how to achieve this with ADFS, as described here, but I haven’t yet seen any guidance for Azure AD.

OK, here’s how to do it.  (Note that this assumes you have already configured the AWS Console to work with Azure AD via SAML)

Go to your Azure Portal and open the Single Sign-On blade for your Amazon Web Services Console application.  Under the User Attributes section, select the checkbox to expose other user attributes, as shown below.

1

 

Select the option to add a new attribute.

2

In the Add attribute blade, set the Name value to “SessionDuration” (note that this tag is case sensitive), the Value to the timeout in seconds that you want, and the Namespace to “https://aws.amazon.com/SAML/Attributes“. Then click OK.

3

The net result should look like this:

4

Save the changes and you are ready to go and test the new timeout.

For more information on the SessionDuration attribute, please see the AWS documentation here:

https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_providers_create_saml_assertions.html

Tony

 

I’ll get to the problem with Powershell Execution Policy shortly, but first a bit of background…

If your AAD/O365 admin accounts are configured for multi-factor authentication (which they should be, because it’s free), you will likely be familiar with the Exchange Online PowerShell Module, which is designed to work with MFA.  Getting to the Module download is not blindingly obvious.  Go to the Hybrid menu option in the Exchange Admin Center and select the second option as shown below.

ps0

 

Once it is downloaded you can launch it and sign-in using the Connect-EXOPSSession command.

OK, that’s all the background.  On with the meat of this article…

Once you’re up an running you might, like me, want to run a script within the session.  This is where things got tricky.  In my case I wanted to run the AnalyzeMoveRequestStats.ps1 script to, well, analyse my mailbox move request statistics as described here.  When I tired to dot source the script as described in the article, I received the standard error you often see when you haven’t got your execution policy set correctly.

ps1

But when I checked my execution policy things looked OK.

ps2

So, what was going on?  After a bit of research, I found there are several different types of execution policy that come into play, as described here.  You can list the current policies by adding the “-list” parameter to the Get-ExecutionPolicy cmdlet.  In my case the current session (Process) was set to RemoteSigned.

ps3

 

The RemoteSigned option was clearly insufficient for my needs, so I had to set it to Unrestricted using the command, Set-ExecutionPolicy -Scope Process -ExecutionPolicy Unrestricted.

ps4

 

After running the command, the ExecutionPolicy for Process now showed as Unrestricted.

ps5

I could now dot source the script.

ps6

Note that you will need to change the execution policy each once per session if you are running scripts in this way with the Exchange Online Powershell Module (MFA version).  There is likely a simpler way to set this permanently, but I quite like the fact that the module re-sets the security each time in this way.  Setting the execution policy to Unrestricted permanently is not a good practice.

Please leave a comment if you know of a different way of achieving the same result – especially if easier ;-)

Tony

I got caught by surprise earlier today when I was looking at some of my older blog posts. It turns out my first entry was on the 10th March 2008. Happy 10th birthday Open a Socket!

Thanks to all of you who have supported me over the years with comments, words of encouragement, and for keeping the verbal abuse to a bare minimum.

Tony

Actually, this is more of a question than answer – although I have an answer of sorts, albeit far from elegant.

I’ve been scheduling some batch onboarding mailbox migrations from a hybrid environment with Exchange 2010 to Exchange Online.  The batch process is pretty straightforward, but I haven’t found an easy way to dump the list of users within the batch after it has been created.  Of course if you are using the CSV import method of populating the batch this is a bit of a non-issue.  However, if you’re creating the batch manually you might want to check back in at some point and remind yourself who is in the batch – or you might want to export the list to send to someone else.  Now I’m not entirely new to Powershell, but I can’t see any obvious way to do this easily.  Here’s what I ended up with:

(Get-MigrationBatch -Identity “MyBatchName” -IncludeReport | select -ExpandProperty report).entries.message.formatparameters | ?{$_ -like “*@*”}

Ugly, isn’t it?  The above command will provide a list of email addresses of the mailboxes within the batch.

If you have found an easier way of doing this, please post a comment below.

 

 

Here’s something I discovered recently and would like to share with you.  If you are using Skype for Business Online and want to control access to it using Conditional Access policy, you should be aware that under certain circumstances the control can be completely bypassed.

The problem has to do with the fact that Conditional Access only kicks-in when the authentication attempt is from the following:

-A web browser
-A client app that uses modern authentication
-Exchange ActiveSync

Conditional Access is not processed by legacy clients, i.e. those that do not support modern authentication.  For example, the Skype for Business 2015 client (the one that ships with Office 2013, and without modern authentication enabled) cannot interpret the Conditional Access policy and as such will bypass the controls.

Let’s look at this in more detail.

In this example, I have created a new Conditional Access policy specifically for Skype for Business Online.

I want all users to be included in the policy.

001

I only want the policy to apply to Skype for Business Online.

002

And finally, I only want access to be permitted from Hybrid Azure AD devices (i.e. those that are joined to on-premises AD and device registered in AAD).

003

I’ve left all of the other settings within the policy at their defaults.  Once I’ve enabled and saved the policy the next thing to do is test whether it works as expected.

The first test is to determine whether the policy blocks access from the Skype for Business 2016 client (click-to-run version) running on a device that does not meet the Hybrid Azure AD condition.  As expected, the access is denied along with a friendly and reasonably helpful error message (shown below).

004

 

The second test is run from a machine that also doesn’t meet the Hybrid Azure AD condition, but this time the sign-in attempt is from the Skype for Business 2015 client.  In this test, the user is able to sign-in without any problems.

005

The Skype for Business 2015 client is effectively able to completely bypass the Conditional Access control, thereby rendering it effectively useless.  Your Skype for Business Online instance can be accessed from any device from anyone who has valid credentials.  The question is then what you can do about it?  There’s currently no silver bullet to handle this scenario.  Microsoft makes provision to block legacy client apps for SharePoint Online and, to an extent Exchange Online, but there is nothing obviously available for Skype for Business Online.

One workaround is to force MFA (at the Azure AD level) for the users that need to access Skype for Business.  With MFA enabled the user sees the following (spectacularly unhelpful) error when trying to sign-in from the Skype for Business 2015 client.

006

I understand that Microsoft are (as of November 2017) looking a method – currently in private preview – to address issues with legacy clients and Conditional Access, not just for Skype for Business, but across the board.  Watch this space.

 

 

 

I recently had a challenge with a customer that had on-premises Skype for Business (SfB) and were looking to migrate to SfB Online. They did not want to federate the two infrastructures, but instead wanted to undertake a re-pointing of users at a given point in time by modifying the DNS records. When they introduced AAD Connect the default synchronisation included the SfB attributes, which is standard behaviour when AAD Connect detects that the schema extensions for SfB are present in on-premises AD. The presence of SfB-related user attribute values in the synchronisation flow caused SfB Online to detect all existing SfB on-premises users as hybrid. It meant my customer could not assign SfB Online access to synchronised users, which would have been a problem for testing the cut-over. The workaround for this was to modify the AAD Connect synchronisation rules to set the SfB attribute values to null.  The steps implemented to achieve this are shown below.

1. Stop the AAD Connect sync scheduler. 

From an elevated Powershell prompt run the following command

Set-ADSyncScheduler -SyncCycleEnabled $false

2. Open the Synchronisation Rules Editor and create an editable copy of the ‘In from AD – User Lync’ inbound synchronisation rule.

sfb1

 

3. Set the new rule to have a higher precedence (lower numeric value) than the original rule. 

sfb2

 

4. Leave the scoping filter as is, i.e. no change.

sfb3

 

5. Leave the join rules as is, i.e. no change.

sfb4

 

6. Edit the transformation for each of the shown values.  Change the flow type to Expression and the source to Authoritative Null.

sfb5

 

7. Save the rule.

8. Start the AAD Connect Sync scheduler and run a full (initial) synchronisation by running the following Powershell commands:

Set-ADSyncScheduler -SyncCycleEnabled $true
Start-ADSyncSyncCycle -PolicyType Intial

9. Confirm that the synchronised users no longer appear as hybrid users in SfB Online.  Run the following Powershell command:

Get-CsOnlineUser | ft userprincipalname, interpretedusertype -AutoSize

Note. This command requires the Skype for Business Online Windows Powershell Module, available here.

The output should show your synchronised users with an InterpretedUserType of  ‘NoService’.  If any appear as ‘HybridOnPrem’ then the custom synchronisation rule has not taken effect.

The synchronised users should now be available to enable for Skype for Business Online.

 

Hopefully, this has been useful to you.  Let me know if you have any corrections or suggestions for improvements by adding a comment.

 

Last week I came across an issue when attempting to create a new custom synchronisation rule in Azure AD Connect. When I tried to finish the wizard and add the rule, I received the error: “Object reference not set to an instance of an object”.

object_reference_not

The workaround is to add a tag to the rule on the Description page, as shown below.  The tag doesn’t need to be meaningful.

tag

The issue appears to be specific to version 1.1.561.0 (July 2017) of AAD Connect. It wasn’t a problem in previous versions and it has been addressed in version 1.1.614.0 (September 2017).

For the latest of AAD Connect versions and version history, see https://docs.microsoft.com/en-us/azure/active-directory/connect/active-directory-aadconnect-version-history

 

I hit this problem while working with Azure AD Connect at a customer earlier this week.  The situation was that AAD Connect had already been configured with Pass-Through Authentication, which was working as expected.  The next step was to enable Seamless Single Sign-On, but this failed with the following: ‘Failed to create single sign-on secret for True’.

sso_fail

The trace log file showed this:

[ 18] [ERROR] EnableDesktopSsoTask: Failed to create desktopsso secret for myadds.org. Exception message: System.Management.Automation.CmdletInvocationException: Exception has been thrown by the target of an invocation. —> System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. —> System.Runtime.InteropServices.COMException: The RPC server is unavailable. (Exception from HRESULT: 0x800706BA)

I immediately thought this was probably an issue with the firewall between the server running AAD Connect and the on-premises AD Domain Controllers.  The difficulty I had was that my customer’s environment was really locked down and I didn’t have access to the firewall or to the tools I would normally use to troubleshoot something like this. So, basically I decided to try and reproduce it with my own Azure AD environment.

As expected, the problem came down to a missing firewall port (TCP 445).  This port is (as of 3rd August 2017) not listed in the port requirements on this web site (Table 1):

https://docs.microsoft.com/en-us/azure/active-directory/connect/active-directory-aadconnect-ports

In my first test, I configured the firewall rules a per Table 1, i.e. without port 445 and attempted to enable SSO via AAD Connect.  I received the same ‘The RPC server is unavailable’ error.

I then removed SSO manually using the Powershell method.

sso_fail_ps1

In my second test, I tried to enable SSO using the Powershell method.  Again, I see the RPC error.

sso_fail_ps2

Looking at the firewall logs, I see a failed connection attempt to the ADDS DC on port 445.

2017-08-02 13:09:47 DROP TCP 192.168.237.130 192.168.237.101 50423 445 0 – 0 0 0 – - – SEND

After adding port 445 to the allowed firewall ports, I re-attempted disabling + enabling SSO using the Powershell method.  This time the cmdlet completes successfully.

sso_fail_ps3

The status now looked good and confirms that the missing firewall rule allowing traffic from the AAD Connect server to the DCs on port 445 (TCP) was the culprit.

sso_fail_ps4

Interestingly, it looks like I would have hit the missing TCP 445 problem when enabling Pass-Through Authentication (i.e. prior to enabling Seamless SSO), but for the fact that I chose to use an existing AD Forest Account.  If I had tried to let the Wizard create the account required for PTA it would also have failed due to TCP 445 not being available.

Hopefully this should give you enough to go on if you come across a similar issue.