WDS Error ESX 3.5 U4
It’s been sometime since I’ve blogged about a technical issue I’ve had to overcome but seeing as I couldn’t find an answer anywhere on the Internet for this one, I thought it best if I shared a little.
I have a WDS Server which is used to compile the build for our Windows Server 2008 environment, however we don’t utilise PXE and so a Media deployment point is used for the majority of installations.
Yesterday I went about updating the unattended.xml using the Windows Automated Installation Kit (WAIK) and validated the file before updating the deployment point. I’m still developing the Windows Server 2008 SOE and wouldn’t be updating things once they were in production. When I started to deploy an image, using a Datastore ISO on a new ESX 3.5 Update 4 Server, as the change occurred from the Windows Boot PE environment to the installation phase I received the following error.
Hitting OK I was presented with the failed deployment wizard screen. and the error: Non-zero return code executing command “X:\Deploy\Tools\X64\TsmBootstrap.exe” /env:SAStart, rc = –214746259.
Firstly I went for the easy solution www.google.com and was only able to find a single reference to the error on Ben Hunter’s BDD Blog and the fix listed is to re-import source files on the deployment point. This didn’t seem like the issue I had, as the install.wim was injected to MDT directly from media and hadn’t been modified.
I decided to take a look through a few of the log files that WDS creates. This can be done from within the Windows PE environment which the installation had returned too. I fired up notepad from the command line browsed to the log path X:\MININT\SMSOSD\OSDLOGS, changing the file type to “All Files” displays all the log files in the common dialog. I thought I’d start with the BDD.log.
No errors in BDD.log and the last command run was the format of the system disk. Next I opened LiteTouch.log and the last line is “<![LOG[About to run command: “X:\Deploy\Tools\X64\TsmBootstrap.exe” /env":SAStart]LOG]!> and time/date stamp. I was getting closer as this was the command the returned a non zero exit code.
On a side topic an exit code of zero indicates that the application or command ran successfully, thus a non-zero means that an error has occurred.
Opening ZTIValidate.log the first line caught my eye. <![LOG[The task sequencer log is located at X:\windows\TEMP\SMSTSLog\SMSTS.log. For task sequence failures, please consult this log.]
Opening the SMSTS.log I can see some errors about being unable to write logs, then I see the C:\ get formatted, so I assume as the disk hadn’t been formatted when trying to write the logs before caused these errors. Sure enough following the format the log is able to be saved. As I continue through the file I can see the steps from the Task Sequence successfully completing, “Inject Drivers”, “Apply Patches” etc.. then I find this error.
<![LOG[Failed to run the action: Install Operating System. A device attached to the system is not functioning. (Error: 0000001F; Source: Windows)]LOG]!>
Since this is a VM I figure isolating the hardware should be fairly quick. To assist in my troubleshooting I decide it would be easier to get the logs of the server and onto my workstation where I could search them easier so I jumped back to the command prompt and checked I had an IP address.
Hmm a little strange, then I remembered that I had selected “Enhanced vmxnet” as the type of Network Adapter when creating the VM. That had me thinking that perhaps I should have selected E1000 or built the server without the adapter at all (at least initially).
Taking the easy option I simply “unconnected” the network card and rebooted to restart the deployment. It once again failed so I completely removed the NIC and it completed with no problems. Once the deployment was complete I simply added the NIC. Now things are working, and moving forward I’ll try and get the drivers injected into PE.