Installing MinShell on Windows Server 2012 Core

Inspired by this great post on improvements in Windows Server 2012 I decided to give MinShell a go.

Unlike that post, however, I was starting from Server Core and wanted to add on MinShell (not start with Server regular and remove the additional components to be left with just MinShell). Not a problem, PowerShell can be used to install the “Graphical Management Tools and Infrastructure” feature.

Hmm, bummer. That was unexpected!

In retrospect this is not surprising. It happens because (obviously) the binaries for the Server Manager GUI bits are not present in the installed system and so one must point the Add-WindowsFeature cmdlet to a source where it can get these from.

Had it been a non-GUI feature such as DNS or DHCP, the binaries would already be installed as part of Server 2012 Core – but not enabled – and so all the cmdlet needed to do would be to enable the feature. (However: if it were a non-GUI feature whose binaries were removed via the Uninstall-WindowsFeature cmdlet with the –Remove switch, then we would have had to point Add-WindowsFeature to a source if we wanted to install these features later).

The Add-WindowsFeature cmdlet has a –Source switch which lets you specify a path from where it can pick up the binaries. This must point to the c:Windows folder and that’s where the binaries are copied over from. If you have another Windows Server 2012 (non-Core) you can share its c:Windows folder and point –Source to that. Or you can mount the Windows Server 2012 (non-Core) install image from WIM file in the installation DVD and point –Source to that. If you have the WIM file from the installation DVD on a network location, you can also point to the Windows Server 2012 (non-Core) install image in that by prefixing the path with WIM: like this: –Switch wim:pathtowimfile.wim:2 (where 2 is the index of the image in this WIM file).

WIM files are sort of like containers for disk images. A single WIM file can contain many disk images – often with overlapping files – and so the size of the WIM file isn’t necessarily equal to the size of all the disk images combined. It is also a file based disk image format – meaning the images contain the actual file system and files within, not just binary data like a sector based format (such as ISO or VHD files). This is why you can actually “mount” an image from the Windows Server 2012 DVD on a folder and then browse this image as a regular file system – because it is a regular file system! The Windows installation DVDs usually contain the installation WIM file – called install.wim – in the Sources folder.

The DISM (Deployment Image Servicing and Management) command is your friend when it comes to WIM files. We met DISM earlier in the context of enabling/ disabling features and while it has a pretty straight-forward syntax it’s kind of ugly. Apart from enabling/ disabling features, you can also use DISM to inspect WIM files, mount the images within them, make changes to these images, and so on.

Before mounting an image in a WIM file, we need to inspect it to see what images are present. The /getwiminfo switch is your friend for that:

As you can see, there are four images in this WIM file. Two of these images are 7GB in size, two are nearly 12GB in size. So one would expect the WIM file containing these images to be around (7*2+12*2=)36GB in size, but in reality the WIM file is just 3.1GB. This is because all these images share the files within them and in addition to that the files are compressed. We want to install the “Graphical Management Tools and Infrastructure” feature which is present in the Windows Server 2012 Standard image (index : 2) and so that’s the image we are interested in.

To mount this image, create a folder and invoke DISM with the /mount-wim switch:

Mind you, DISM is very finicky when it comes to syntax. There is no space between the /wimfile: switch and its arguments; similarly no space between /mountdir and its arguments.

If we navigate to the c:\Offline folder we can see a regular file system with a Windows folder. Navigate through this and you’ll see its just like the Windows folder in a live Windows system. This is the folder we are interested in.

So back to Add-WindowsFeature we now pass the above mentioned folder via the -Source switch:

Success! Restart, and type servermanager in the command prompt that you get to launch MinShell!

DISM mounted images are persistent across reboots. So once you are done with it be sure to unmount the image with the /unmount-wim switch:

The /discard switch tells DISM we are not interested in it saving any of our changes to the image (not that we made any changes here; but in theory one could and save it back to the image).

Have fun with MinShell!