‘The security database on the server does not have a computer account for this workstation trust relationship’ when trying to log on via RDP to a customized VM in Azure.

Last summer, our customized lab VM’s started having issues after updating from Windows Update and applying the latest fixes.

The error message received was ‘The security database on the server does not have a computer account for this workstation trust relationship’ when trying to log on via RDP.

After extensive troubleshooting we determined that this seems to have been caused by changes which renamed the VM internally, either in Azure Resource Manager, the Windows Update code or the version of the Azure IaaS agent running on the VM’s.

At any rate… as the server was a domain controller, when it renamed itself to match the external name of the VM it caused an identity crisis as it could no longer find itself.

Yes… the technically correct answer from the handbook is to Sysprep/Generalize any image/template VM that you deploy to Azure but for a lab scenario you sometimes want to have a specialized image that contains AD & Friends – and AD and Sysprep are just not friends as Sysprep will eat AD and not even leave a skeleton behind.

Prior to June this year, deploying a customized non-sysprepped VM from an image with a new name to Azure Resource Manager did not change the actual computer name inside the image if the VM was a domain controller.

This meant you could deploy the same lab image multiple times to Azure with a different external VM name but the name of the VM inside the OS would stay the same.

At some point between June and August this year there seems to have been a change to this which was affecting the image, after updating from Windows Update the name of the server inside the VM was now being changed to match the external name.

This in turn resulted in it being impossible to log on to the VM after rebooting, as the actual computer name of the domain controller VM in the AD that it hosts isn’t changed to the new name at the same time.

TLDR: Redeploy the VM using the original VM name, run Windows Update and update without problems.  Create a new image from the updated VHD as a new template VHD.

I’ve tested a couple of different deployments after updating the specialized image and it doesn’t look like there is an issue updating from Windows Update using the new template that contains the latest updates, even if deployed again using a different name.  Either the rename is dependent on a specific component being updated or possibly it’s related to the version of the Azure agent running on the VM.

The Domino Effect: ‘extensible-connector-refresh-required’

Symptoms: MIM Domino MA run operations return the error message ‘extensible-connector-refresh-required’

Cause: Updating the Domino MA requires the COM interface on it to be refreshed

Resolution: Open the Domino MA in the MIM Sync MMC and press OK, the MA will start refreshing the COM interface (and the user will need to press OK 4 more times as the MA cycles through different sections for updates).

Lesson: The Domino MA doesn’t have a ‘Refresh Interface’ button like a standard normal ECMA2 would, if you open the MA configuration and press OK rather than Cancel it detects the refresh requirements and initiates a refresh.




Parsing JSON files with PowerShell on Azure VM’s using ConvertFrom-JSON

The Outdated Newtonsoft.JsonDLL Conspiracy

The Outdated Newtonsoft.Json.DLL Conspiracy

From our special agent Json Bourne: notes from the field.

Make sure you are reading a raw string: i.e. (Get-Content JsonFile.JSON -raw) | ConvertFrom-Json

PowerShell seems to use an outdated Newtonsoft.Json.DLL file which doesn’t fully support comments

Make sure there are no comments in the .JSON file – if there are any you may need to manually edit them out to get ConvertFrom-Json to behave

The two versions being used on Azure VM’s seem to be (October/2013) and (February/2014).

Support for comments in JSON files in Newtonsoft.Json.DLL was first added in version 5.03 from 2014/11 (support for single line comments) and expanded on in later versions (6.07+)

The latest version of Newtonsoft.Json.DLL is 10.0.2 (as of April/2017)


Newtonsoft.JSON on Github

If you put a comment into a Json file and try to run ConvertFrom-JSON in PowerShell on it you will get the following error message:

ConvertFrom-Json : Invalid object passed in, ‘:’ or ‘}’ expected. (1): {
At line:1 char:28
+ Get-Content .\config.JSON |ConvertFrom-Json -Verbose
+                            ~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo          : NotSpecified: (:) [ConvertFrom-Json], ArgumentException
+ FullyQualifiedErrorId : System.ArgumentException,Microsoft.PowerShell.Commands.ConvertFromJsonCommand


An Introduction to JavaScript Object Notation (JSON) in JavaScript and .NET (February 2007)





Installing MIM 2016 on a named SQL instance with a non-standard port


FIM 2010 and MIM 2016 have a thing in common with regards to SQL named instances; they can’t handle the truth.  In this case they can’t handle that the SQL instance is running on a port other than the default 1433 SQL port.

What do you do when you have have a friend that can’t handle the truth but that you don’t want to lie to?  You tell them as much of the truth as they can handle and that’s exactly what we need to do for MIM.

In this case, truth comes in the form of an SQL alias, your friend MIM will never know the truth that it’s actually talking to SQL on a port other than 1433.  Shocking!

Note: The SQL alias needs to be created in both the 32-bit and 64-bit versions, your 32-bit applications will use the 32-bit alias while the 64-bit applications use the 64-bit one (obviously).
If one application works with the alias while another one doesn’t its usually an indicator that there’s a difference between the 32-bit and 64-bit aliases in your setup.

For testing and troubleshooting purposes it is useful to create an empty file with a .udl extension, for example test.udl, and then double-click that file to launch it.

A Test .udl fileudl2

Make sure you’re testing the same provider that your application uses, the current version of the MIM Setup program seems to use the SQL OLE DB provider rather than the SQL Server Native Client.

You also need to make sure that the application you’re testing with supports the cryptographic ciphers and algorithms being used on the server end, for example; if the application requires a version of TLS that is disabled on the server side then you need to either get an update to the application or enable the cipher on the server side.

Continue reading

How to administer AzureAD, O365 and Skype for Business using PowerShell and Multi-Factor Authentication

Azure Active DirectoryPreviously, support for MFA in O365/AzureAD/Skype/Sharepoint was limited to Office applications that supported it and browser-based administration of O365/Azure.

This changes with version 1.1 of the Azure AD PowerShell module released earlier this month which provides support for MFA.


The steps to enable it are as follows:

  1. Enable MFA on the various tenants ()
  2. Download the latest AzureAD PowerShell modules that provide support for MFA (v 1.1 released 15. August 2016)
  3. Make sure you have the correct mobile phone number, alternate email and/or authenticator app installed (you typically want to have more than one MFA option available)
  4. Enable MFA on the user you want to protect with MFA
  5. Instruct the user to go to https://account.activedirectory.windowsazure.com/profile/ to verify their MFA settings (and SSPR if applicable)

Note: There are still some aspects in the Windows OS that are still not really aware of MFA, particulartly the Domain Join functionality.  If have enabled MFA on the account you’re using for the domain join operation and you receive an erroneous “Incorrect Password” error (i.e. code 0x52e in the NetSetup.log debug log)  during a domain join (and assuming you are actually typing in the correct password) then you may need to revert back to using a separate non-MFA account – at least for the domain join operation.

Default MFA settings:

The Office 365 tenant/resource host (Exchange Online, SharePoint Online and Skype for Business Online) will need to be configured to accept a modern authentication connection.
Here is the per service state of modern authentication by default :

  • Exchange Online – OFF by default.
  • SharePoint Online – ON by default.
  • Skype for Business Online – OFF by default.

Once you have MFA enabled and the new version of the AAD PS module installed you should be able to go through the additional MFA verification steps after logon:MFA


Azure Active Directory PowerShell with Modern Authentication

Download Details: Azure Active Directory Connection

Azure AD PowerShell: Public Preview of support for Azure MFA + new Device Management Commands

Azure AD PowerShell: Public Preview of support for Azure MFA + new Device Management Commands

How to install and configure Azure PowerShell

Skype for Business Online: Enable your tenant for modern authentication

Exchange Online: How to enable your tenant for modern authentication

The sign in experience with Azure Multi-Factor Authentication

Remote Desktop Gateway and Azure Multi-Factor Authentication Server using RADIUS

What are App Passwords in Azure Multi-Factor Authentication?


Integrating SAP Web Services with MIM – part1


I recently finished a successful integration project for a customer where they were moving from FIM 2010 to MIM 2016 and from the legacy SAP Management Agent to the current SAP Web Services Management Agent – the following are observations from the journey.

The default projects published for the SAP Web Services Connector contain VB for Workflows code, rehosted so you don’t need Visual Studio to edit it but the drawback is that you’re limited to editing it inside the accompanying Web Services Configuration Tool (which is a bit….dated, frankly).  Documentation for it is also very limited and hasn’t been updated since the original release of the client.

The files discovery.xml and generated.config (or cfg.config in the older version) inside the .wsconfig project can be edited using 7-Zip or similar if you need to tweak settings that currently can’t be configured inside the WSConfig tool, such as increasing the HTTP timeout value of the Web Services Connector, the default timeout being 60 seconds (standard Caveat Emptor warnings apply to modifying the content).

Note that the generated.dll file inside the .wsconfig project is re-compiled when you make changes to the Discovery settings inside the WSCONFIG tool, which in turn overwrites any manual changes made to the generated.config file so you’ll need to redo the manual changes whenever a change to the Discovery settings in the project is made.

Also, make sure you have downloaded the latest SAP Web Services Connector from Microsoft, see the Wiki on MIM Management Agents.

Be aware that since the latest versions of the Web Services connector changed build numbers from 5.3.x.x to 1.x.x.x you may need to uninstall older versions of the SAP Web Services Connector since the Windows Installer logic will reject the lower build version numbers as it (quite logically) believes it to be an older version of the connector.

For the Employee object in SAP, the following Standard SAP BAPI functions are being referenced in the pseudo-VB code and the generated.dll file that live inside the .wsconfig project that comes with the SAP Web Services Connector:


Full Import:


The latest SAP Web Services Connector template (2016-03-02) expands the number of SAP object types to also include pseudo-VB code for the following object types:

 _OrganizationUnit ()

Full Import:

PrimaryGroup ()

Full Import:

Export Replace:

 SecondaryGroup ()
Full Import:

Export Replace:

Profile ()
Full Import:

Export Replace:

Role ()
Full Import:

Export Replace:

User ()
Full Import:

Export Add:

Export Delete:

Export Replace:



Importing using the Standard BAPI calls from SAP and the default project from Microsoft turned out to be quite slow as we discovered after the initial deployment.  The default project is using the standard BAPI calls to parse through a list of all users in the alphabet to import one entry from SAP at a time – which is probably part of the performance issue.

Fortunately, the local SAP-consultant I was working with on this project had written a custom BAPI which was collecting the users being imported into a table of rows and then sending it over the wire in one batch – we ended up dumping all the VB-code from Microsoft from the original default Web Services project and only using it for reference in a new web service project I wrote specifically to use the new custom BAPI.

In short; The performance boost from using the custom BAPI to communicate with SAP Web Services was staggering – an Import operation from SAP Web Services to the MIM Connector Space that took 6 hours with the standard BAPI calls now takes ~3 minutes using the custom BAPI.  I’ll go further into this in part 2 of this series.


Further details:

BAPIs – Definition

MIM and SAP Web Service Connector Timeouts

Integrate SAP HR and Active Directory using Forefront Identity Manager (FIM) SAP Connector for WS


ADFS SSO Primers

AD FSCloud, Office 365
SSO means different things to different people – make sure you’re talking about the same thing (i.e. Single Sign-On, Simple Sign-On or Same Sign-On)
In order to transparently obtain a claim from ADFS the following must be present:
  1. Your browser or client application needs to support WIA and claims
    IE/Edge support WIA out of the box, Firefox can reportedly be configured to support it (Chrome and others possibly as well using the same steps)
  2. Your browser must be talking to the intranet ADFS servers (as talking to the ADFS proxies takes you through forms-based authentication)
    This is frequently done through either a split-brain DNS setup or a network device that routes internal claim requests directly to the internal ADFS servers and external claim requests to the ADFS proxies
  3. Your browser must trust that the ADFS farm is an intranet web site that (the site should be in the Local Intranet sites in Internet Settings)
    Adding a GPO setting to include the ADFS farm name will work, targeting the GPO at the Computer gives the User the flexibility to add any required additional entries as the user settings inherit any sites added at the computer settings level.
  4. You must be logged on to a domain-joined machine as the user you expect SSO to be provided for
    Note that a cached logon doesn’t really count as a proper logon (you’re simply being allowed access to local resources on the workstation based on a previously successful logon)
  5. The Relying Party application should redirect you to the ADFS farm when it detects you are attempting to log on using an account beloging to a domain suffix that their IDP is responsible for
    This should preferably done immediately upon entering the account name when switching focus to the password field, entering the password at the RP side shouldn’t be required at all (f.x. an onFocus event targeting the password field, similar to what https://portal.office.com does).



Support for Windows Integrated Authentication in Firefox

Configuring Chrome and Firefox for Windows Integrated Authentication

onFocus event

Split-Brain DNS

A faulty split-brain DNS configuration can prevent a seamless SSO sign-in experience

Office 365 Single Sign-On with AD FS 2.0 whitepaper

A federated user is prompted unexpectedly to enter their credentials when they access an Office 365 resource

A federated user is repeatedly prompted for credentials when he or she connects to the AD FS 2.0 service endpoint during Office 365 sign-in

Quickly undeleting users, groups or Organizational Units using the AD Recycle Bin feature

Note: if the AD Recycle Bin isn’t already enabled in your forest then you need to dig out your last system state backup and go through the restore process – enabling the ADRB does not help with recovering any objects deleted prior to it being enabled.  See this article for enabling it. 

A VM snapshot can potentially be used to restore from for DR purposes but only if *both* the virtualization host platform and the underlying guest OS supports it (W2k12 or later for the guest OS and Hyper-V W2k12 or vSphere/ESXi 5.0 Update 2 for the virtualization hosts). 

If your VM snapshot doesn’t fulfil the minimum OS and virtualization requirements then you may still be able to use it as a last resort but the operation will be a lengthy process if the snapshot DC was brought online before manually setting the “Database Restored from Backup” registry key on it to force it to generate a new invocation ID.

From the VMware article on VM-GenerationID support:

VM-Generation ID functionality requires the hypervisor vendor to create the virtual machine identifier and expose it to the guest. VMware has provided this functionality in the following releases of vSphere:

  • VMware vSphere 5.0 Update 2 (vCenter Server and ESXi must both be at 5.0 Update 2)
  • VMware vSphere 5.1 (ESXi must be at least 5.0 Update 2)

From the Technet article on Safe Domain Controller Virtualization:
Beginning with Windows Server 2012, AD DS virtual domain controllers hosted on hypervisor platforms that expose an identifier called VM-Generation ID can detect and employ necessary safety measures to protect the AD DS environment if the virtual machine is rolled back in time by the application of a VM snapshot. The VM-GenerationID design uses a hypervisor-vendor independent mechanism to expose this identifier in the address space of the guest virtual machine, so the safe virtualization experience is consistently available of any hypervisor that supports VM-GenerationID. This identifier can be sampled by services and applications running inside the virtual machine to detect if a virtual machine has been rolled back in time.
Virtualized domain controller cloning relies on the hypervisor platform to expose an identifier called VM-Generation ID to detect creation of a virtual machine. AD DS initially stores the value of this identifier in its database (NTDS.DIT) during domain controller promotion. When the virtual machine boots up, the current value of the VM-Generation ID from the virtual machine is compared against the value in the database. If the two values are different, the domain controller resets the Invocation ID and discards the RID pool, thereby preventing USN re-use or the potential creation of duplicate security-principals.


Let´s look at the AD Recycle Bin in action in a lab:

First, we check whether the AD Recycle Bin feature is enabled in my lab forest using the following Powershell command:

get-ADOptionalFeature “Recycle Bin Feature”


If the EnabledScopes part is empty it means the AD Recycle Bin still hasn’t been enabled in your AD forest, luckily I already have it enabled in my lab.

Now, I just deleted a few items from my AD lab, some of which shouldn’t have been deleted and now need to be restored quickly.  I know all the deleted objects had ‘Test’ in the name so let’s see what shows up in the AD Recycle Bin when searching for objects with the isDeleted AD attribute set to ‘true’ and that include Test somewhere in the name and then sorting the list by the AD attributes msds-lastKnownRdn and lastknownParent to see where they were originally deleted from:


Of these 6 items, I’m only interested in restoring the group objects (the other ones were intentionally deleted).  I’ll add a filter to only target groups and then pipe it to the Restore-ADObject Powershell command with the -WhatIf switch appended to determine what action would be performed on which objects without actually performing the restore:


Great, the -WhatIf option shows me the command would restore only the two groups that I want to restore and leave everything else we don’t want to restore still in a deleted state in the AD Recycle Bin.  We can now confidently run the same command again without the -WhatIf switch (Adding the -Verbose option instead to report the status for each restore):


Excellent! The two groups and their pre-deletion membership state are now back in business without having to bother with a lengthy restore process!  Let’s confirm the groups are really back in the correct OU in AD by running Get-ADObject again without the ‘isDeleted’ filter option:

The undeleted groups are back where they belong in AD with the correct membership in the minimum amount of time required to recover from the original error.

This will of course *not* replace your regular backup procedures but it should really be considered a mandatory precaution to enable it in any AD environment running Windows 2008 R2 DC’s as it really cuts down the time required to recover from accidental deletion of objects in AD.

In W2k12 R2 there is now also the option of using the Active Directory Administrative Center to undelete objects if you want a more graphical way of doing things:
Additional links:

AD Recycle Bin Best Practices

AD Recycle Bin behaviour

Safe virtualization of Domain Controllers

vSphere versions required for VM-Generation ID support

Fixing When Your Domain Traveled Back In Time, the Great System Time Rollback to the Year 2000

Virtualized Domain Controller Architecture

Things to consider when enabling the AD Recycle Bin

Recycle Bin

#1 check if you have any applications that use the DirSync control and make sure you have applied the NTDS hotfix from KB979214 to Domain Controllers (if you are running Windows 2008 R2 DC’s) as this changes the default behaviour of the relevant LDAP search control when returning values for objects that reference recycled (soft deleted) objects.

#2 read the ASKDS article on best practices for implementing the AD Recycle Bin.

#3 figure out if you have any critical applications using the DirSync control that can’t handle the isRecycled attribute state for objects that have been soft deleted.


ILM 2007 will unfortunately not be able to handle the AD Recycle Bin but FIM 2010 with Update 1 will (ILM 2007 being out of support for more than a few years now should also really be reason enough to upgrade to at least FIM 2010 R2 or MIM 2016).

Applications that have the potential to be impacted negatively by enabling the AD Recycle Bin are most likely going to be older product versions that focus on directory synchronization or monitoring using the Dirsync control.  Those will either need to be retired or updated to a newer version (older GALSync and MIIS instances probably fall under the same unsupportable hat as ILM 2007).


Additional links:

Polling for AD Changes using the DirSync control

FIM 2010 Update 1

The DirSync control search does not return the deactivated linked attributes from a modified object in a Windows Server 2008 R2-based domain

Using the DirSync Control

PowerShell DirSync Sample