Windows server 2012 r2 Hyper-V installation of a role or how errors can be made during installation

A small background to the question I want to tell. In our organization, the Internet is distributed through usergate - a maximum of 40 people at a time. In the past six months or a year, the problem has worsened. More and more often a page appears in the Internet browser, which says that you need to stand in line for the Internet.

Two years ago I did a test version of a proxy server. Made it on Zentyal 3.3, via squid. I went on vacation, I came - the test car was gone. Since the problem was not so acute, they scored about it for a while, and besides, there was simply nothing to build a new proxy on.

At the end of the year before last, we had a new server under 1C with powerful hardware, on which a virtual machine could be spinning, and at the same time, not one.

I read repeatedly on Habré how to put a role of Hyper-V, how to configure a network. However, there were few useful publications. Why? So I'll show you now.

On January 30 of this year, he decided to put the Hyper-V role on the server. On the same day, I put the technical preview of Windows 10 on my working machine in order to see what was there and how, and put Hyper-V on my machine. But was upset: Intel Core 2 Duo E8400 does not support Hyper-V virtual machines.

This was the first "cant": not on every computer where you can put Windows 8 and 10, you can start the hypervisor.

Attempts to install ClearOS from a USB stick onto a personal machine were unsuccessful.

All this inspired me to install a hypervisor on an accounting server.

Started installing the role, selected one network interface cardof the two physical only. This was another "jamb": as it turned out later, not the last, but the most important.

In this choice, only one virtual network switch was created for the selected physical network card. The hypervisor did not allow after installation to create another virtual switch, did not allow to put into operation through the existing virtual switch, created for one physical network card, the second network card.

To be simpler, when a hypervisor is installed and you have two or more physical network cards, you all need to select them to create virtual switches. What for? When you create a virtual switch, the settings from the physical network card disappear, the card can be used simultaneously by several virtual machines.

In my case, the data was copied from only one network card, while the second, just like the first, stopped working as usual. The role of Hyper-V in the network card settings can be removed, but it will be removed for the remaining network cards.

The virtual switch is turned on and off only with a tick, turning it on again by disabling this switch in hidden devices leads to a blue screen of death. Another achivka - caused a blue screen of death on the server 2012.
You can also remove it only through the virtual switch manager, other methods of removal lead to problems.

It was my next “cant”, after removing the Hyper-V role from the network card, everything became as it was before the installation of the hypervisor. But I could no longer connect to the Virtual Switch Manager. It was clear that somewhere in the registry all this hangs and does not allow to recreate everything again.

Another "cant": deleted the virtual machines that used the virtual switch. The hypervisor has stopped working. I registered the settings in the network card - but the server from this also stopped working normally. I couldn’t access it in the shared folder, I couldn’t access the network from it. And again, dancing with a tambourine: most likely, it was not an included “client for Microsoft networks”, but now I don’t want to check it. As well as making these mistakes, it’s specially just to take screenshots of the screen.

After that, I was already thinking of restoring the Windows server through the recovery points. But I decided to try reinstalling the role of the hypervisor.

Where so many stocks come from? Everything predetermined the wrong choice of network interface cards. It is necessary to choose all physical ones and most likely not physical ones either (another “jamb”). When a virtual switch works, it becomes a layer between programs and hardware. But since I chose only one, and the virtual switch actually put the second network card out of operation (oops, I don’t remember exactly whether I managed to work with the second). The virtual machines did not see the local network through an external network card and this made me dance with a tambourine.

After removing the role of the hypervisor, I reinstalled it. But I already selected all the physical network cards and made a mistake again. I think that a Hamachi virtual network adapter should also be chosen. Now, after installing two virtual switches for each physical network card from the virtual machine, I see the network from the virtual machine, and the virtual machine is visible from the network.


If you have chosen only one network interface card, and you have two or more of them, then it is better not to dance with a tambourine, trying to connect through the internal network interface card and nat. Make it simpler - remove and install the role of the hypervisor again, so you save yourself nerve cells and time, which, like the cells of course. Not every processor is suitable for lifting a virtual machine on an embedded hypervisor.

PS ClearOs on the virtual machine did not start normally - version 6.5 i386, some kind of problem with the image output at boot time.

Also popular now: