locked
Online Storage Migration causes loss of source vhd's RRS feed

  • Question

  • 7 node hyper-v cluster r2 sp1. Using vmm 2008 R2 sp1.
    I did a storage migration of a running vm that had multiple vhd's.  Each vhd had its own csv.
    So the config was OS vhd attached to ide controller and all the other vhd's attached to the scsi controller.
    I only changed the location of the vm configuration and the vm os vhd for the storage migration.  All the other vhds locations remained unchanged.  I was not moving the storage to another host, just moving the storage to a different csv.
    The migration was successful to step 1.8 (export completed successfully which is 1.7).
    At this point the migration failed with 2903 cannot find a particular avhd.
    I then ran a repair > undo action in VMM. This failed with:
    Error (12700)
    VMM cannot complete the Hyper-V operation on the vm-n01 server because of the error: 'Win08 Server' failed to remove snapshot. (Virtual machine ID 418D4208-9A86-4B02-81C7-A560C90DB593)
    Cannot delete snapshot: The system cannot find the file specified. (0x80070002). Snapshot ID AD1D0002-E14B-4FF7-ACA7-FA4866DC6216. 
     (Unknown error (0x8003)) 
    I then tried a repair > undo action again and received the same 2903 error.
    At this point I noticed that all the source vhd's in all the csv's were gone.  The only remaining source vhd was the OS vhd.  The only file appearing in all the other csv's was a avhd file.
    So, my only option at this point was to start the vm using the original source vhd and restore all the other vhd's from backup.
    I did open a call with MS support and they stated a restore was the solution as well.

    • Edited by chireader Thursday, September 15, 2011 8:33 PM
    Thursday, September 15, 2011 8:32 PM

Answers

  • The were no prior checkpoints before running the migrate storage job. The migrate storage job created two checkpoints.

    The main question is why the OS vhd associated with the ide controller remained with its avhd and all the other vhd's associated with the scsi controller were deleted leaving only the avhd in its place.

    Its my guess the job failed due to some of the csv's having low free space.  But this should only resort in a failure not data loss.  It seems the prereq check when the job starts needs to be a little more aggressive with its checks.

    Friday, September 16, 2011 3:21 PM

All replies

  • Did you have any checkpoints? If so, were these checkpoints merged outside of VMM (i.e. snapshots merged in Hyper-V manager?
    Steve Thomas, SSEE, Microsoft
    App-V/MED-V/SCVMM/SCCM/AppCompat
    http://madvirtualizer.wordpress.com/
    The App-V Team blog: http://blogs.technet.com/appv/
    The MED-V Team Blog: http://blogs.technet.com/medv
    The SCVMM Team blog: http://blogs.technet.com/scvmm/

    “This posting is provided "AS IS" with no warranties, and confers no rights. User assumes all risks.”
    Thursday, September 15, 2011 11:58 PM
  • The were no prior checkpoints before running the migrate storage job. The migrate storage job created two checkpoints.

    The main question is why the OS vhd associated with the ide controller remained with its avhd and all the other vhd's associated with the scsi controller were deleted leaving only the avhd in its place.

    Its my guess the job failed due to some of the csv's having low free space.  But this should only resort in a failure not data loss.  It seems the prereq check when the job starts needs to be a little more aggressive with its checks.

    Friday, September 16, 2011 3:21 PM