I brought this issue up ages ago, I thought we addressed it, but maybe we didn't since it only shows up in odd/rare scenarios.
The issue is caused by upgrading the OS, a timing problem, and a relationship design. After an OS version upgrade, if Windows Server 2008 discovery runs before Windows Server 2003 discovery does, then we have a short period of two OS objects. Eventually, OS 2003 discovery runs and "undiscovers" the 2003 computer, OS, etc. However, because of the way logical disk hosting relationship is created, the 2003 logical disks are not deleted when their hosting parent is deleted, because the 2008 discovered an instance of the same type for them to be hosted. To work around this problem, whenever you do an OS version upgrade, uninstall the agent from the server being upgraded (already required) and DELETE the agent from SCOM. Then - when the OS version is completed - re-push the agent to the new OS version and only 2008 objects will be discovered. To recover from the broken state, simply DELETE the agent from SCOM. This will mark all discovered objects as "deleted" in the database, then let the agent reconnect and approve it from pending. Only the current version objects will be discovered and the 2003 disks will stay deleted. From: [email protected] [mailto:[email protected]] On Behalf Of Tyrell, Tommy A. Sent: Tuesday, December 16, 2014 6:54 AM To: [email protected] Subject: [msmom] Windows Server discovery problem I have multiple Windows servers in my SCOM 2012 R2 management group that are discovered as both Windows 2003 and Windows 2008 servers. Specifically the Windows logical disk monitors for both Windows 2003 and Windows 2008 are enabled for these servers producing duplicate alerts. The servers were originally installed with Windows 2003 but then upgraded to Windows 2008. I can't get the Windows 2003 logical disk monitors to "un-discovery" these now Windows 2008 servers. Any help would be appreciated.
