We came across a problem when trying to implement SCOM 2012 agents for a client with satellite sites connected via slow links.  The agent would flip-flop into the “Not Monitored” state and it was difficult to determine why.  We assumed it was a networking issue and focused our time with the network team.  But some changes to Active Directory may be necessary….

Scenario

For this client, the satellite sites are  on poor links back to HQ.  But the interesting aspect is that the satellite sites move – as they are large ships.  The ships have a self-sufficient infrastructure for when the ships go out of range.  Meaning many servers and a local Active Directory child domain.

Issue

We found that both SCOM gateway and SCOM agents installed but they greyed out often.  We checked the network which was slow but “fine”.  As the SCOM Agent appeared to sporadically connect it seemed likely the problem was due to the movement of the ship and the effect on the network,.

We  posted on TechNet and with others assistance we found there were some timing issues which prevented SCOM agents from working.

Resolution

Some very helpful information and scripts allowed us to replicate the issue and show that the latency for Active Directory actions were the cause of the problem.  We need to install a domain controller representing the remote site within the same location as the Management Server.  We are in the middle of testing this but thanks to all in the SCOM community for assisting us as always.

We will update this post once we have the full results.

If you need to un-install the SCCM 2007 agent you may find Software Centre remains in the Program Files directory and there isn’t an option in Control Panel to help you finish the process.

In this scenario you can run the CCMsetup.exe again but with the uninstall switch. on the affected server:

1. Open cmd prompt as administrator, run the following commands:

2. Cd c:\windows\ccmsetup

3. ccmsetup.exe /uninstall

Wait 5 minutes and then you should be able to confirm that software centre has been uninstalled and you will see it has disappeared from the Start Menu.

Now delete the following files and folders :

• C:\windows\CCM
• C:\windows\ccmsetup
• C:\windows\SMSCFG.INI

The SMSCFG file is important as this stores agent connection details to your old SMS/SCCM server  Hope that’s useful..

I decided to share a few discoveries I made along the way to doing the re-install of SCCM 2007 R3 on a Windows 2003 server after attempting to do so in a client’s test lab.

Things to consider when you remove an existing SCCM 2007 R3 installation…

First, ensure this is the only option left available to you, if you run a multi level hierarchy that supports a large client base then re-installation is probably not an option.

If you remove a working installation of SCCM (or pretty much any application) you will find that other services may still be installed – WSUS and MDT spring to mind. It’s worth keeping this in mind for when you install the core SCCM services again – more likely that you will need to update these products after the re-installation.

It’s also worth keeping in mind that SCCM will install and update some common OS files and components, you may have newer versions of files and registry keys on the serve that  will cause warnings to appear – warnings that can be addressed once the product is re-installed and re-patched.

SMS Provider Fails on Install
A fairly common issue with a re-install, the install process fails at the SMS provider component. If you spend time looking at other websites the common items to check are some of the following:

• Check to see if the Firewall is enabled – http://technet.microsoft.com/en-us/library/dd547071.aspx
• Ensure that the SPN is set – http://technet.microsoft.com/en-us/library/bb735885.aspx
• Check to see what permissions the machine account has on the SQL server and if it is a member of the local admin group.
• Check to make sure that that install account has the correct SQL Database Permission.
• The discovery class leftover from the old installation Root\NetworkModel is removed http://configurationmanagerblog.blogspot.co.uk/2011/07/setup-failed-to-install-sms-provider.html

If the above does not succeed, try looking at the ConfigMgrSetup logfile and also the MOFCOMP.LOG file.  In my exprience something similar to this may appear:

An error occurred while processing item 2 defined on lines 21 – 62 in file C:\Program Files\Microsoft Configuration Manager\bin\i386\NetDisc.mof: Error Number: 0×80041026, Facility: WMI

Prafull Wavhal points out the information about the leftover class, which was a great starting point but additional information in the log files file may indicate there could be an issue with the registration of a certain class itself.

If you suspect this, compare the  RTM version of this file with the SP2 version to see if there were any subtle differences between lines 21-62 as indicated in the error.  Something like this may be missing:

boolean bAMTDiscovery = false;

As part of the installation of SCCM 2007 SP1 the class is added but is not removed when you remove SCCM itself – even when we manually removed the Discovery class earlier. Simply adding this line into the existing NetDisc.mof file or replacing the file with the SP2 version will allow the MOF to be compiled and the installation will proceed.

No SCCM Components are Installed

The installation of SCCM proceeded but at the point where the install completed the SMS components were in an unknown state and the basic components were in a “install pending” state. SCCM says that the install has completed but taking a look in the event viewer a number of errors are reported around the installation of the components themselves:

SMS Server 1014: Component SMS_SITE_COMPONENT_MANAGER on computer COMPUTERNAME reported: SMS Site Component Manager is installing this component on this site system.

LoadPerf 3012: The performance strings in the Performance registry value is corrupted when process SMS_SCHEDULER extension counter provider. The BaseIndex value from the Performance registry is the first DWORD in the Data section, LastCounter value is the second DWORD in the Data section, and LastHelp value is the third DWORD in the Data section.

LoadPerf 3009: Installing the performance counter strings for service SMS_SCHEDULER (SMS_SCHEDULER) failed. The Error code is the first DWORD in Data Section

SMS Server 1016: Component SMS_SITE_COMPONENT_MANAGER on computer COMPUTERNAME reported: SMS Site Component Manager failed to install this component on this site system.

If you take a look at the Site Status inside the SCCM Configuration Manager console you may not see any components visible and no site systems. You may not also be able to launch the ConfigMgr Service Manager with an error about not being able to contact the Site Server.

At the point where the install of the components is launched, the Perfmon counters cannot be installed due to an underlying issue with the Performance Counters. In this case, the following command will perform a reset of the counters.

lodctr /r: c:\windows\system32\perfstringbackup.ini

To continue restart the SMS_COMPONENT_MANAGER service and the components should proceed to install.

Note: I also noticed the issue with components when I did a site reset, the reset performs a reinstall of the components and the same errors were reported – more than likely due to having an existing issue with Performance Counters more than anything particular.