Best practices for mass reimaging? Having unmanaged duplicate clients with mismatched Resource IDs. (2012 R2)

Question

We reimage about 5000-8000 clients each summer. Last summer we were still at RTM, but have since moved on to R2. We build images (Win7 SP1) in vsphere and capture via capture media TS. Last year's image didn't get SCCM client uninstalled prior to capture
so we would have random issues with computers going to unmanaged status and some that would show up without a client cert. To avoid this issue we stripped the client out prior to capture this year.
I believe this is how we handled the reimage process last year as well, but I am not positive on that. We were also dealing with a lot of new laptops last summer, where as they obviously have existing records this summer. Since SCCM replaces the wired MAC
address with the wireless MAC (laptops) we can't just toss these into an OSD collection because it won't pickup the OSD advertisement / PXE. (Is there any workaround for this?) Since this is the case, we are blowing away each client's AD account and DDR in
SCCM, then doing mass import of hostname and wired MAC into SCCM, dumping them into the appropriate OSD collection, and they image unless they happen to pickup last year's PXE deployment that first has to be cleared, or unless they had a motherboard replaced
and our MAC database didn't get updated. We did the mass import a week ago and the manual entries are listed with hostname and MAC and entry date of 7/9/2014. This week we started imaging. Almost immediately after reimaging (at which time the AD record is
created upon rejoining domain) we see a second account show up in SCCM from AD Discovery with dates of 7/14/2014 and 7/15/2014. Neither account is managed or shows that SCCM client is installed, but it shows a site code.
The manual entry lists an agent name, agent site, wired MAC, name, NetBIOS name, Resource ID of 167xxxxx, assigned site, and CCM GUID.
The AD Discovery record shows agent name, agent site, domain, IPv4 and IPv6 addresses, name, NetBIOS name, primary group ID, domain, Resource ID of 20971xxxxx, resource name and type, SID, assigned sites, container name, UAC, etc.
Why won't these records merge and show up as being properly managed? I am not yet sure if they fix themselves after one record or the other is deleted. Obviously this process isn't working well and it removes the clients from their direct membership collections
and AD groups. I'd think that all of this could be avoided if we just had the wired MAC persist in the DDR.

Answer 1

Last year's image didn't get SCCM client uninstalled prior to capture so we would have random issues with computers going to unmanaged status and some that would show up without a client cert. To avoid this issue we stripped the client out prior to capture.
There is no reason or need to do this. There is no correlation between the two as long as the client agent was properly prepared (which does happen with capture media although you should strongly consider using a build and capture task sequence). Clients
are perfectly capable of living within an image -- I do it all the time and it is a common practice.
Since SCCM replaces the wired MAC address with the wireless MAC (laptops) we can't just toss these into an OSD collection because it won't pickup the OSD advertisement / PXE. (Is there any workaround for this?)
This is not correct and thus also unnecessary as ConfigMgr will use the MAC Address *or* the SMSBIOS GUID of the system to determine targeting during OSD. The SMBIOS GUID is an immutable unique ID set by the OEM that is part of the resource record in ConfigMgr
also.
Jason | http://blog.configmgrftw.com