KB2982791 Also Appears to Impact WDS Multicast Deployments

Question

I had a client who ended up installing KB2982791 on one of their deployment servers. This server is running Windows 2008 R2 WDS and MDT 2012 Update 1.
Once this was installed, the following behaviors were observed:
Server would hang on the shut down screen
Multicast Deployments of a Windows 7 Enterprise custom image would still deploy with a low number of clients.
Once the number of clients simultaneously connecting increased to an undetermined number however, the MDT task sequence would start to error out with "Network Name Not Found" or "could not find ZTIBootstrap.wsf" among other error messages.
We saw the effect at around 20 machines in various stages of the task sequence, but it might have been less when the problem exhibited itself.
It looked to be some type of memory or handle leak, as any client still multicasting the image would continue to do so albeit at a GREATLY reduced transfer rate. My hunch is that at some point they would error out altogether.
When the deployments on the clients was interrupted and restarted, they would hang on the "Attempting Multicast Transfer" portion of the task sequence. It was as if the WDS server was no longer participating in the multicast group. Performance
indicators however were nominal for the system.
If we rebooted the server, operation returned to normal and we were able to restart the task sequence. The problem would recur however once the number of simultaneous deployments increased to a high level (something more than 10, but definitely at 20 we
would see the issue).
We uninstalled 2982721 as recommended and the problem immediately went away. The server was able to shut down correctly, and we no longer saw clients throwing errors.
We realize of course that KB2982721 has been pulled, but just wanted to throw out there another issue we saw with it that I don't think has been documented thus far. I'd be interested to hear if anyone else has seen other weird effects or if they feel there's
a different potential cause of the issue we observed.
Thanks,
Rick Reuling
Walker IT Group, LLC

Answer 1

Unfortunately after further testing the problem reappeared, so at this point we don't feel the patch was to blame just a timing coincidence.
We were able to trace the problem to a specific notebook - whenever it joined the multicast transfer all the other machines would stop receiving and everything would hang up. We suspect a bad NIC or motherboard.