none
Importing new drivers caused deployment to fail - what to do next? RRS feed

  • Question

  • We are using the 'total chaos' method of Out-of-box drivers in MDT, and our workstations are mainly Lenovo laptops and Dell desktops.

    Recently, we added drivers for a new Dell desktop.  The method was as per previous, (1) download the cab of Win 7 drivers, extract (2) Run the 'unblock-file' powershell on the driver directory to remove the 'downloaded from internet' flag (3) Import into the Windows 7 folder in Out-of-box-drivers (4) Regenerate Boot Images and update the boot images in WDS.

    The deployments then started failing, both on the new desktop model and on all the existing models.  We see the image applying, unattend.xml applying, and then we see this error.

    Windows failed to start.....

    File: \Windows\system32\drivers\amdkmpfd.sys

    Status: 0xc0000359

    Info: Windows failed to load because a critical system driver is missing, or corrupt.

    Since I can't get to a command prompt, I'm not sure where I can find deployment logs.

    I recreated the boot image fully, in case it had gotten corrupt, and it reported some errors during the import of drivers.  Perhaps these had also happened the first time, I hadn't been watching the output as closely.


    Deployment Image Servicing and Management tool
    Version: 6.2.9200.16384

    Image Version: 6.2.9200.16384

    Found 1 driver package(s) to install.
    Installing 1 of 1 - D:\AdminDeploymentShare\Out-of-box Drivers\Net\netwex_15.6.1.6_58FA89356668673D4CCB913FDC4A27CCE0F04A19\netwex.inf: Error - An error occurred. The driver package could not be installed. 

    For more information, check for log files in the <windir>\inf folder of the target image.

    Error: 2

    The system cannot find the file specified.

    The DISM log file can be found at C:\Windows\Logs\DISM\dism.log

    Exit code = 2

    DISM /Add-Driver failed, rc = 2

    The dism.log is massive and I don't know which bits to post. 

    I don't know how (or if) I can pull out the drivers for the new model and return things to a working state.

    The nuclear option is to restore a version of the deployment server from before when these drivers were added, but that can be awkward.

    Another option might be to re-organise all the drivers, away from Total Chaos, and towards Total Control, if indeed that might be the root cause of the problem.  What do you think is the best option?

    Friday, April 11, 2014 11:14 AM

Answers

  • From the error:

    # for hex 0xc0000359 / decimal -1073740967 :
      STATUS_INVALID_IMAGE_WIN_32                                   ntstatus.h
    # The specified image file did not have the correct format,
    # it appears to be a 32-bit Windows image.
    # 1 matches found for "0xc0000359"

    Is it possible that the file is corrupted? Perhaps you are attempting to load a 64-bit driver on a 32-bit OS?

    Boot Back into WInPE and grab the bdd.log file. then grep for all instances of "xcopy" in the bdd.log file. this will represent the list of drivers *actually* copied down to the local machine. Find the culprit and try removing. Specifically, was this drivers *supposed* to be installed on this machine?  Can it be installed later on in the deployment process (perhaps as a application install?)

    Sadily I have encountered drivers before that "don't play well with others": http://keithga.wordpress.com/2013/12/09/mdtosd-driver-management-overview/


    Keith Garner - keithga.wordpress.com

    Tuesday, April 15, 2014 7:08 AM
    Moderator

All replies