There’s one thing that I can’t get enough of when automating Windows deployments, it’s ambiguous and confusing error messages. More please, I’m a sucker for punishment.

Here’s one I’ve been troubleshooting on and off for a while - an unattended Windows deployment via MDT fails with error 0x80004005, which is about as helpful as a brick life vest.

Similarly, the MDT log files are pretty useless in narrowing down a root cause. Whilst I couldn’t use the log files to help me discover the issue, I have been using MDT to deliver the same task sequence on multiple hypervisors in our lab environment.

As it turns out the issue can be caused by a fairly specific configuration - delivering Windows Server 2008 R2 with the Remote Desktop Services Session Host (RDSH) role enabled via UNATTEND.XML to a virtual machine on Hyper-V with Dynamic Memory enabled and the Startup RAM set to 512MB.

Dynamic Memory is probably something you would not run for RDSH in production, so there’s a few workarounds:

  1. Set the Startup RAM higher than 512MB
  2. Don’t use Dynamic Memory for an RDSH server (the recommend solution)
  3. Use PowerShell to reconfigure the Dynamic Memory settings after Windows has been deployed to the virtual machine