Network bridging in VM broke connectivity to host; what did it do?

Question

I screwed something up royally, but I want to understand how exactly I screwed it up. Bear with me because I'm not a networking expert.

Here is how my network is configured:

     [HyperVHost]
     157.57.109.75 /24, gateway 157.57.109.1
     10.245.7.1    /24

     [VM]
     157.57.109.137/24, gateway 157.57.109.1
     10.245.7.57   /24

What I was trying to do is upgrade a virtual machine using a third-party VHD. I was having difficulty getting this VM to obtain Internet access, even though it looked like the networking setup was the same as the old one, as far as I could tell. Thinking that maybe the fact that there was two network adapters was to blame, I made a desperate attempt and horrid mistake; I bridged the two connections.

Suddenly, I lost my remote desktop connection to HyperVHost, permanently.

I don't have physical access to this machine, so I'll probably have to ashamedly ask for help. What I'm wondering is what exactly bridging does and why bridging the connections on VM took down the connection for HyperVHost as well.

Answer 1

It is possible, that by bridging two connections You have created a bridging loop. Such loop would flood the network, by infinitely repeating all passing packets - including e.g. ARP replies of gateway.

The switch could block port to prevent massive DoS of local network.

HyperVisor would definitely DoS himself with zillions of pps. Machine with bridging loop dies in matter of seconds.

Answer 2

Bridging an interface is turning the interface into a switch where many hypervisors will act as a router and put the guests behind a nat. I'm not familiar with HyperVHost but when you are setting up bridging on a linux box its a two step process: change the physical interface to accept bridging and second add a virtual (logical) interface that is bound/bridged off of the physical interface for the host to use. VMs are then able to have layer 2 connectivity to the network they are attached to instead of having a nat/routing layer in the middle.

My guess is you likely did the first half of that process but now the machine doesn't have a logical interface configured for layer 3.

If you have access to a computer on the same subnet you could try running an arp command to see if the interface's mac address is showing up. Cisco switches have a show arp command as well.

Answer 3

If memory serves what happens when you bridge the NICs under Windows the OS creates a new virtual NIC and uses this for connections. If this is done remotely it can (and usually does) server the remote connection because the old IP address is just gone.

Microsoft recommends that Hyper-V have a dedicated NIC to use for networking to help prevent this type of problem. There is a setting when you add the NIC to hyper-v networking to allow the host to use the NIC for non-hyper-v activities (i.e. remote desktop, etc.). That should prevent a situation like this in the future. see 7 Best Practices for Physical Servers Hosting Hyper-V Roles

Someone will need to go hands-on with the host however to break the bridge.