Windows 2012 R2 deployement failes on physical machines

Hi,

I'm struggling to get W2K12R2 deployed via OSD on my physical DELL machines.

I've captured an image (via VM) with SCCM 2012 R2 CU1. When I deploy the image to a VM, it is working fine. When I'm deploying it to a physical machine, it fails. During tasksequence some roles are added via dism, but dism failed because the image is corrupted (not valid package). After applying the image and reboot, the image gets corrupted.

I have included all patches in the image (before capture) that were available including the rollup of november.

Could this be driver related?

Could it be related to the patches?

Regards,

WiM

January 8th, 2015 4:22pm

What error do you get? Was do the status messages and smsts.log tell?
Free Windows Admin Tool Kit Click here and download it now
January 8th, 2015 4:36pm

How did you capture your image, on a virtual or physical machine?
January 8th, 2015 4:58pm

Are you deploying to physical servers? If so it could be driver related. Remember that you also need to consider drivers for RAID controllers. Dell have provided the Server Deployment Pack (which integrates with ConfigMgr) for this very purpose.

http://en.community.dell.com/techcenter/b/techcenter/archive/2013/06/26/dell-server-deployment-pack-2-1-for-microsoft-system-center-configuration-manager-is-available-now

Free Windows Admin Tool Kit Click here and download it now
January 8th, 2015 5:05pm

First, as Torsten requested, we need to know actual technical details here otherwise everything we say is just a guess. Specifically, what task is failing and what does smsts.log say about the task.

Also, as a side note, did you build your image from the Microsoft provided image that includes the November update? If not, I'd highly recommend you rebuild your image using the Microsoft image with this update already slipstreamed into it.

January 8th, 2015 5:08pm

Well,

I did capture the image on a VM

I am deploying it to physical DELL servers

The task sequence fails at the moment that features of roles are added via tasksequence, because dism failes. So that is the error that you see in the SMSTS log (and in the dism log). If I open a command box during tasksequence, and launch dism /online /get-features, I get an error: "The specified package is not valid Windows package". So the error message is not the cause but the result of the image already being corrup

Can it be related to drivers at this stage? I also tried without applying drivers but did not make any difference.

I did use the image of october 2014. Should I download a new image again? Indeed a lot of patches are available for this image...

Thanks,

WiM 

Free Windows Admin Tool Kit Click here and download it now
January 8th, 2015 8:08pm

I absolutely would build my image using the latest one released from Microsoft.

I don't think your image is corrupt. What features exactly are you trying to add and how are you trying to add them? Can you please post the entire, unedited, relevant snippet of the smsts.log?

January 8th, 2015 8:13pm

Just to check for understanding:
The same TS (with the DISM commands) works when deployed to a VM?

And just a shot in the dark: Do you have a reboot before DISM, and the reboot is configured to boot to the boot Image?

Free Windows Admin Tool Kit Click here and download it now
January 9th, 2015 1:52pm

I was about to ask exactly the same question as Martin above:

  • Is the machine in FullOS or in WinPE during the DISM command?

You state that you open CMD and run "dism /online /get-features" and get a failure, this sounds like you're running the dism command in WinPE instead of FullOS.

You could change your dism command to "dism /image:%OSDisk% /get-features", where %OSDisk% is of course the drive letter of your OS (which is a good to store in to a variable like OSDisk during the Format & Partition -step). With this in mind, you could either change the machine to boot to FullOS before running the DISM commands you're running or use the Offline -method and run the commands in WinPE. I've noticed that when these commands are run from WinPE, they do run a little bit faster. I usually run them just after the Apply OS -part and in WinPE.

January 11th, 2015 6:33am

I was trying to install APPL server role

I was executing the dism command after the server restarted with installed os, not WinPE and yes, on VM's everything was working.

I followed Jason's advice and downloaded the latest DVD (and captured the image not via SCCM but imageX).

I don't know if the DVD was the cause, but it seems to work now. If I have time, I will capture the image again via sccm capture media to exclude this as a possible cause, which I doubt.

BTW, is there a way to update existing W2012 machines to R2 without manually starting the setup on the server?

thanks

Free Windows Admin Tool Kit Click here and download it now
January 11th, 2015 11:30pm

BTW, is there a way to update existing W2012 machines to R2 without manually starting the setup on the server?

To my knowledge, no. There might be a commandline parameters for setup.exe though.

January 12th, 2015 4:17am

I don't know if the DVD was the cause, but it seems to work now. If I have time, I will capture the image again via sccm capture media to exclude this as a possible cause, which I doubt.


I had similar problem with dism adding features from SXS folder, and root cause was bad DVD media.
Free Windows Admin Tool Kit Click here and download it now
January 12th, 2015 9:42am

It seems that you can run the setup.exe with an unattend answer file. I'm trying to create it with WSIM... but not succeeded yet...
January 14th, 2015 2:30pm

What exactly are you trying to do? Server 2012 to Server 2012R2 is a complete upgrade of the OS, not just a swap out or addition of a few components.

There's no way to upgrade (in-place) to 2012 R2 without running setup.exe. Yes, you can automate the process using an unattend.xml file, but what's the goal here and how is it related to the original thread?

Free Windows Admin Tool Kit Click here and download it now
January 14th, 2015 3:19pm

You're right, maybe I should start another thread.

For me it is related because I have to deliver new machines and update existing ones, but it is not the correct forum indeed for the second part.
What am I trying to do: upgrade windows2012R2 without human intervention (because we have a large number of machines running the exact same config and we want to bring them to R2, because the new delivered servers will also be R2)

January 14th, 2015 3:28pm

OK, so technically, this is *not* supported by ConfigMgr although this is kind of a grey area.

Microsoft upgraded a large portion of their internal clients from Windows 8 to 8.1 (equivalent to 2012 to 2012 R2) using an in-place upgrade deployed by ConfigMgr. The details are all at http://www.microsoft.com/en-us/download/details.aspx?id=40811 and http://www.microsoft.com/en-us/download/details.aspx?id=41965.

You may be able to do the same with Server 2012. I doubt it would be explicitly supported. Not knowing business goals for doing this, I would say it's not worth the effort in the first place to do the upgrade at all, but that's not for me to decide.

Free Windows Admin Tool Kit Click here and download it now
January 14th, 2015 3:48pm

Bad news... back in the same situation... Our 2012 R2 deployements are failing again.

After the image is applied, the server is rebooted into the installed OS... when typing dism /online /get-features we get the same error again:

Error:0x800f0805
The specified package is not valid Windows package.

In smstslog, there are no errors while applying the image, only after reboot we see that the image is corrupt. This results in the fact that dism fails, no pachtes are visible and setup.log is corrupt... I already tried to restore health via dism, it states that errors are fixes but nothing gets fixed.

I tested for a week on physical machines without any issues (with the image captured from november DVD). On friday 16/01 we started deploying VM's again because everything seemed fine, but the 2 new VM's failed. For some odd reason, our test VM's still succeed (which already exist for a long time)!

After we saw the VM's failing, we tried the physical machines again... which also failed then.

We have created new VM's with the exact same config as the test vm's but can't figure out for what reason the OSD failes or succeeds.
We moved our DP's to other ESXes, moved our VM's around, but there was no consistency in the fail/no fail.

The first time I succeeded to install a new VM was when  I removed the VNIC of a failing VM, and added a new one again.

Last week, I again created a new package with the same captured image, again it was working. Now, after the weekend, it is failing again.

I think in total we did about 50 installations in different scenarios, but still no clue what is going wrong. It feels like something is going wrong on the network....

Time to open a case at Microsoft?

January 27th, 2015 4:28pm

Download the latest R2 media, import install.wim to your ConfigMgr. Do build and capture task sequence using the .wim and deploy .wim you built.
Free Windows Admin Tool Kit Click here and download it now
January 27th, 2015 6:52pm

Narcoticoo,

Just to be sure, the other methods are supported and should work, no? (Tested via capture media and also via imageX, both with same result)

Is there a specific reason why you suggest this method? The problem is that it will work again for a unknown period... and then fail again... for me it seems impossible that the image is corrupt, because it should fail from the first time.

the only thing we install before the capture is dot net 3.5 and patches...

January 28th, 2015 8:07pm

Because it's the standard way of doing things (unless you want to use MDT to build your images). Could you post your build & capture and deployment task sequences in whole here?

Free Windows Admin Tool Kit Click here and download it now
January 29th, 2015 5:04am

We are suspecting that the OSversion in vmware from the reference machine had something to do with it (using vmware Vsphere client and vsphere web client). Web client was showing correct version (2012), but the Vsphere client was showing 2008. We created a new VM with correct version to build the reference machine, since then we didn't face this issue anymore

Thanks


  • Edited by WiM8 19 hours 13 minutes ago
April 24th, 2015 8:33am

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics