When a backup job using snapshot technology (like DPM) runs against Server 2012 Hyper-V, NTFS might briefly report free volume size to be 0 for a brief period. If VMMS queries free volume size and find that there is none it will pause VMs. So if VMMS is unlucky and does the query at the moment when NTFS reports 0, the Virtual Machines will go into a paused state. Unfortunately they do not unpause when the space frees up.

In our environment we experienced this with a CSV volume with 5.7TB free out of 10TB. One of our VM’s went into a paused State because Windows Server 2012 thought that it had run out of space. At first I thought this was because it was performing some type of calculation relating to the potential of dynamic disks.

This issue is being reported here:

http://social.technet.microsoft.com/Forums/en-US/winserverClustering/thread/7a7ddaca-fb09-4c9a-b0ae-9f3b5e10694d

and here:

http://social.technet.microsoft.com/Forums/en-US/winserverhyperv/thread/d9ed7259-104b-4b7f-8e57-2cc5364f5381

KB article on the topic:

There is now a hotfix available but you officially need to contact Microsoft to obtain it. The hotfix is KB2791729.
I’m posting it here for those in an emergency situation. While I have not modified this in any way, I am not liable for any damages that this hotfix causes in your environment, so use at your own risk.
https://skydrive.live.com/redir?resid=349D2CCB95BB5F15!164&authkey=!AEt2WGm00Gsx0N0