1

My Situation:

VMWare ESXi 5.0 SSH Access

One of my VMFS volumes is completely full. The VMFS volume has 1 thick provisioned (lazy zero) disk with 1 snapshot on it.

On one of my other VMFS volumes I have enough space for this disk but only if I were to thin provision it.

The VM is off and will remain off until the disk in question is repaired.

My Solution:

I want to use vmkfstools to clone the disk (deleting the snapshot in the process) to the VMFS volume.

vmkfstools -i "/vmfs/volumes/Datastore1/myvm/myvm-0001.vmdk" "/vmfs/volumes/Datastore2/myvm/myvm.vmdk" -d thin

I will then delete the current disk with the snapshot from the vm and connect the newly created thin disk with the same SCSI connector in it's place.

My question is when it is deleting the snapshot does it create a helper snapshot and if so does it place it on the source disk or the destination disk?

2
  • It seems to me that addressing the underlying issue of not having enough disk space would be a better approach to solving this problem. Can you add an extent to the current datastore?
    – joeqwerty
    Oct 23, 2012 at 17:45
  • I'd prefer to safely get rid of the snapshot and leave the storage alone. Nothing is planned for that storage and the snapshot shouldn't have been taken or left hanging to create this problem. Oct 23, 2012 at 17:50

2 Answers 2

0

Your best bet can be to get/create some additional disk storage.

One solution can be NFS.

  • Set up an NFS server.
  • Move the files, disk images there
  • when the size is reduced you can move it back to its final place.
1
  • This is the current plan. I was hoping to clone the disk to another datastore just out of sheer speed. Copying 1.7 TB over the network is making me look for a quicker alternative. Oct 23, 2012 at 19:29
0

I have tested this now and I can confirm that a VMDK with a snapshot can be cloned to another datastore without the creation of any additional files on the source datastore. Also when you select to do a thick to thin conversion during the clone it will only use the space actually allocated by the VM in the source VMDK. I did this while the source VMDK was not being actively accessed by any VM.

This seems to be a viable alternative to the snapshots ate my hard drive space problem when you don't have enough space to copy a flat VMDK to another datastore but you do have enough space for the VMDK if it was thin provisioned.

Naturally this is not recommended or entirely safe but it does give an alternative solution if needed.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .