ConfigMgr wake-up proxy and port security

FYI, got the below message from one of our network guys. This involved another location, not us (thankfully!). Bottom line, if you're implementing port security, do NOT use ConfigMgr wake-up proxy:

We ran into a really “interesting” issue this week.  The SCCM guys decided to implement the SCCM “wake-proxy” feature on the SCCM.  Not quite the normal magic packet WOL procedure.  This has the SCCM server hit specific machines, which then go and hit the machines in their vlan.  However, these “manager machines” also spoof the mac addresses of the machines that go to sleep in their vlan in order to keep the MAC alive in the switch tables.

This is quite fun when you have port-security turned on that limits the number of mac addresses on the ports!  We had random machines dropping off the network, DHCP issues, etc.  It was quite fun to chase down what exactly was going on.  The only indication that we saw was that ports were showing multiple MAC addresses when we knew that there was only one device connected to it and it was not running a VM.  Wasn’t even logged into.

Fortunately, something ticked the back of my mind about a feature they were thinking about implementing on the SCCM and we were able to put the pieces together.  Otherwise, we would still be scratching our heads over this.

Bottom line?  DO NOT implement SCCM wake-proxy with any sort of port-security enabled.  Better yet, just don’t implement wake-proxy at all.

Here is a discussion we found afterwards, once we knew what to look for.
https://supportforums.cisco.com/discussion/11835361/mac-address-flapping-and-sccm-wake-proxy

I worked with ALU TAC all afternoon on this.  The only weirdness that we could see was that the packet capture showed that the computer was sending out replies to ARP packets that were not for his MAC address.

“Cannot edit the object, which is in use by ‘username’ at Site ‘configmgr site’

If you've ever been in the middle of editing an application and had the ConfigMgr console crash, you've likely seen this message.

myitforum.com has a quick and easy way to unlock the object so you can get back to editing! Article is linked below, but here are the two queries to run in SQL Server Management Studio:

This query gets the ID of the locked object:
select * from SEDO_LockState where LockStateID <> 0

This query will delete the locked object, allowing you to get access again:
DELETE from SEDO_LockState where LockID = ‘<LockID of the record identified in the previous query>’

http://myitforum.com/myitforumwp/2013/02/22/unlocking-configmgr-2012-objects/

Computer in console not showing client as installed

We found a handful of computers that had the ConfigMgr client installed, but the console was not reflecting this. After some investigation on an affected computer, I noticed the client certificate said "none":

This is not normal. In addition, some of the items under the Actions tab were missing. Fortunately, I found this blog post that contained the solution. 

The computer was stuck in provisioning mode. The simple fix was to change the HKLM\SOFTWARE\Microsoft\CCM\CcmExec\ProvisioingMode registry key to "false", then reinstall the ConfigMgr client.  SEE UPDATE BELOW

NOTE: The safest bet is to reimage the system in question. If the step that removes the computer from provisioning mode didn't work, it's quite possible there were other issues during imaging. But this is a quick fix if you can't reimage for some reason.

H/T Daniel Classon

UPDATE: Thank you to Microsoft MVP and all-around swell guy Nash Pherson for pointing out that the registry method does not entirely remove the computer from provisioning mode. This article shows how to do it properly via a WMI method:

Invoke-WmiMethod -Namespace root\CCM -Class SMS_Client -Name SetClientProvisioningMode -ArgumentList $false

Or, run the following from an elevated command prompt:

powershell Invoke-WmiMethod -Namespace root\CCM -Class SMS_Client -Name SetClientProvisioningMode -ArgumentList $false

Gotcha - setting registry values via ConfigMgr

We have a package with a program that is supposed to set a registry value. This is a simple command, right?

reg add HKLM\Software\Whatever /v "KeyName" /t REG_SZ /d "Value" /f

When we deployed the program, the value would not be where we expected it, instead we found it here:

HKLM\SOFTWARE\Wow6432Node\Whatever

This blog post explains why this happens (32-bit process running on a 64-bit OS).

The fix is easy, simply add "%windir%\SysNative\" in front of your REG ADD command, like this:

%windir%\SysNative\reg add HKLM\Software\Whatever /v "KeyName" /t REG_SZ /d "Value" /f

Works like a charm!

H/T Brpo's Blog

PowerShell App Deployment Toolkit - "Launching this application has been temporarily blocked..."

I am a huge fan of the PowerShell App Deployment Toolkit. It allows us to deploy or upgrade applications that require certain applications to be closed in a friendly manner. Instead of just killing a process (say, Internet Explorer) before running a Java upgrade, you can let the user know that that IE needs to be closed and let them finish any unsaved work before closing it. There are numerous other wonderful features of this utility, I encourage you to read more at the link above.

That said, we had a strange experience with a recent upgrade from Java 1.7 to 1.8. A number of users were unable to launch Internet Explorer after the upgrade. They all had this error message, which is a window from the PS App Toolkit:

A little Googling led me to this article, which suggested a registry key may be the problem, Indeed it was:

The full value of the key was wscript.exe "C:\Users\Public\PSAppDeployToolkit\AppDeployToolkit_BlockAppExecutionMessage.vbs"

Removing this registry key solved the issue. This is unexpected behavior from this utility, it typically would remove this key upon completion. Perhaps it's a bug in the current version (v3.5.0). Still, it's a problem for us, so we needed to make sure this wouldn't affect any other users.

The solution was to add a line of code in the post-installation section of Deploy-Application.ps1.

Remove-RegistryKey -Key 'HKLM:SOFTWARE\Microsoft\Windows NT\CurrentVersion\Image File Execution Options\iexplore.exe' -Name 'Debugger' -ContinueOnError $True

The full post-installation section is shown below:

##*===============================================
##* POST-INSTALLATION
##*===============================================
[string]$installPhase = 'Post-Installation'

## 

## Clean up IE registry key
Remove-RegistryKey -Key 'HKLM:SOFTWARE\Microsoft\Windows NT\CurrentVersion\Image File Execution Options\iexplore.exe' -Name 'Debugger' -ContinueOnError $True
      
## Display a message at the end of the install
Show-InstallationPrompt -Message 'Installation complete.' -ButtonRightText 'OK' -Icon Information -NoWait

This solved our issue, and clients are upgrading and working as expected.

UPDATE: We still see a handful of computers that experience this issue. We resolve it by pushing them the below PowerShell script. It also adds a registry value in a hive we often use for other things (and we use this for the detection logic in this app), then it notifies the user with a popup window that IE should be working properly.

# Delete registry key
remove-itemproperty -path "HKLM:\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Image File Execution Options\iexplore.exe" -name Debugger

# Create and set value in Chico registry section for detection method purposes
New-ItemProperty -Path HKLM:\SOFTWARE\Chico -Name IEDebugger -PropertyType String -Value "Fixed"

# Notify user
$wshell = New-Object -ComObject Wscript.Shell
$wshell.Popup("Your Internet Explorer issue should be resolved. Please try running it again. If there are still issues, please contact ITSS at x4357.",0,"Done",0x1)

ConfigMgr clients using previous server site code

I've run into a handful of systems with this issue since we migrated to our new CM2012 server. Some clients still have the old server's site code, no matter how we install the new client. The site code value can be found in the registry, and removing the value and installing the new client did the trick for us.

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\SMS\Mobile Client\

Source: http://myitforum.com/myitforumwp/2012/10/05/configmgr-client-gpo-assignment-removal/

Windows not activated after OSD, ConfigMgr client not fully installed

I saw this email the other day on the Windows - Higher Ed mailing list:

I have run into a strange issue that I have not encountered before. I am trying to image a new Dell Precision T1700 desktop with Windows 8.1 using SCCM. I have imported all the drivers from the Dell CAB into CM. The machine images fine, loads software, joins the domain, etc. When I login, it fails to have the CM client fully installed, doesn't see our KMS server for activation, and oddly enough, is missing all the event logs that are normally found under “Applications and Services”. 

The exact same thing had happened to me the day before imaging a Dell OptiPlex 9020. The cause was a recent Microsoft security update that required two restarts. These updates sometimes will completely kill your task sequence, but sometimes will allow the image process to complete, only to find oddities like the emailer mentions above.

An easy workaround is to inject the problem updates directly into your WIM and update the distribution point in ConfigMgr. This way they won't cause any reboots during the task sequence. You can also just inject all updates if you choose (I do, as it makes the imaging process go a bit faster), but I've heard some users say they've had issues with this.

There are a couple ways to do this, one is using the DISM tool, which works great, but is done via command line. The other is to use the built-in ConfigMgr feature, Schedule Updates, which uses a GUI and is really easy. I prefer easy.

This should resolve this issue. Let's hope Microsoft doesn't release any more of these double restarters, but I'm sure they will. You can keep track of the known troublemakers here.

H/T to ConfigMgrBlog (although this is not just a Windows 7 issue)