deleting snapshots causes vmdk files to be created

[AlphaMagus]

I was running out of space on a datastore so I deleted a bunch of snapshots for one of the VMs and now I have even less space. I am using vSphere 4.0 to manage three ESXi 4.0 servers.

Originally, as shown in the "summary" tab of the VM, the provisioned storage was 140 GB with used storage showing 105 GB. I shutdown the VM then I went into snapshot manager and chose to "delete all" snapshots. It took about 12 hours to complete. There were around 15 snapshots deleted and it ran overnight so I wasn't watching it. Now, the VM is showing provisioned storage at 264 GB and used storage at 264 GB. The datastore is showing only 3 MB free.

I browsed the datastore and it looks like there are 7 different .vmdk files numbered from 1 to 7 all around 35 GB in size. I didn't look at the files for this VM in the datastore before deleting the snapshots so I can't say that these files didn't exist before but... WTF? Is it safe to delete them?

Looking in the settings for the VM, the hard disk is showing as vmname-00007.vmdk which is the last one of the bunch. If I clone the VM to a different datastore, will all these files get transferred too? I don't have 264 GB free on the other datastore so I can't just test it. Is this sequence of events typical?

Thanks in advance for any insight into this.

 

[cabraun]

You might just be better off to V2V the VM to a new VM. Those vmname-#####.vmdk files are your snapshots. It looks to me like perhaps not all the snapshots could be deleted before you ran out of space.

 

[Provogeek]

This is a typical problem when snapshots are being used as backups.

The resolution of this issue is a horrible one (IMPO) you will need to grow the size of the datastore. You most likely (though I have not tried) will NOT be able to successfully clone the VM to another datastore. You will (should) run into the same issue as moving a VM with snapshots, you will loose everything since the first snapshot was taken.

So when I had to address similar for a client, the only resolution that did NOT blow up the VM was growing the size of the datastore to give the VM the room it needs for temp files when committing disk changes to the base VMDK on a VM with snapshots.

Please review this best practice for working with snaps-shots:

http://kb.vmware.com/selfservice/mi...n_US&cmd=displayKC&externalId=1025279

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Senior Network Engineer
Keep IT Simple[/color red]

http://www.kiscc.com

Novell Platinum Partner Microsoft Gold Partner
VMWare Enterprise Partner Citrix Gold Partner

 

[AlphaMagus]

Thanks for the help guys. Lesson learned, do NOT use snapshots as a form of backup. I found that the only way to clean up the space used was to manually delete the vmdk files from the datastore. I think that when I chose to delete all snapshots, it tried to merge the data in the snapshots which caused the jump in used hard drive space and then crashed when it ran out of space. When I went to clean out the snapshots of my other VMs, I found that deleting them one at a time worked.