Having an issue when moving Virtual Machine Storage on a Clustered vm is breaking (corrupting?) the vhdx.
Setup:
2-node Hyper-V Server 2012 (the free one) Cluster connected to an EqualLogic PS6100E using a CSV. (2) Teamed (Hyper-V) LAN nics and (3) iSCSI MPIO nics each.
Starting Location:
- VM named "vv-2012" is built from an ISO and running on a server in the node - NOT added to Failover Cluster Manager as a role yet.
- Disk lives at "C:\ClusterStorage\volume1\vv-2012\Virtual Hard Disks\vv-2012.vhdx"
- VM lives at ""C:\ClusterStorage\volume1\vv-2012\Virtual Machines\xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.xml"
- Machine is Running at Login Screen directly after setting the local Administrator password
- Running current Integration Services
Test 1 - From Starting Location - Move VM Storage using Hyper-V Manager - Move just the vhdx one folder up ("C:\ClusterStorage\volume1\vv-2012\vv-2012.vhdx")
- SUCCESS - vhdx moved and machine still running
Test 2 - From Starting Location - Import into Failover Cluster Manager - Move VM Storage using Failover Cluster Manager - Move just the vhdx one folder up ("C:\ClusterStorage\volume1\vv-2012\vv-2012.vhdx")
- FAILED - Disk Read error in the running VM as soon as the move starts (see below image)
Test 3 - From Starting Location - Import into Failover Cluster Manager -SHUT DOWN RUNNING VM - Move VM Storage using Failover Cluster Manager - Move just the vhdx one folder up ("C:\ClusterStorage\volume1\vv-2012\vv-2012.vhdx")
- SUCCESS - Started VM - no errors
Working on the, now, broken vhdx:
- Searching technet (and the Internet - Google/Bing) returns almost exclusively P2V conversions using disk2vhd.exe from Sysinternals. That is NOT the case here - reading these posts talks about partitions not being imported (boot partition) or incorrect
IDE/SCSI settings on imported vhd's.
- Attaching the vhdx to another vm (or my local machine using Disk Management) and running chkdsk returns "Corrupt master file table. Windows will attempt to recover master file table from disk." "Windows cannot recover master file table. CHKDSK
aborted." (also tried with /F even though screenshot does not show that - same result) - I see the 2 partitions (boot and system) but cannot browse them - marked as RAW
- Running TestDisk returns invalid MFT (I think, not an expert on this program).
Summary:
- Moving vm vhdx's that are turned off either through
Hyper-V Manager or Failover Cluster Managerwork fine
- Moving vm vhdx's that are running through Hyper-V Managerwork fine
- Moving vm vhdx's that are running through Failover Cluster Managerbreaks (corrupts?) the vhdx.
So my questions are 3 fold:
- Can anyone else reproduce this?
- Is it a Hyper-V / Cluster bug?
- Can I recover the vhdx?
Wasn't sure if I should post here or Cluster forum? Both?
:) -Andrew
Images: