Datastore Browser, moving vmdk to windows, insane file size

[Volstag]

Greetings all,

I have a situation wherein a virtual machine on ESXi went kaput.  As a result, we're trying to move the *.vmdk off the MSA SAN onto a windows machine for a hopeful conversion to VHD.  If this is posted to the wrong forums, please accept my apologies.

Here's the problem that I'm running into:

I'm using the DataStore browser.  I see two *.vmdk files for the erstwhile virtual machine:

machinename_1.vmdk (103.51 GB)

machinename.vmdk (9.5 GB)

My first question: what's the difference between the *.vmdk and the *_1.vmdk?

I'm using the DataStore browser to download to a 500 GB external hard drive attached to an Windows XP machine.

Here's the problem / situation I'm running into:

First, on the windows side, it creates two files (no matter which of the above two files I download):

machinename_1.vmdk

machinename_1-flat.vmdk

First question: how do these files relate to the two vmdk's as they exist on the SAN?  Is it necessary to download both of these files?

Secondly, and most concerning, the file size shown on the DataStore browser isn't even close to the resultant file size on windows.  For example, I started a download of the machinename_1.vmdk thinking it would be around 103 GB on the windows box.  Unfortunately, each time I've tried this, the file size is way way over the 100 GB stipulated.  The current download I have going now has already downloaded 114 GB and the DataStore progress bar seems to imply it's only about a 1/3rd of the way done.  What am I not understanding here?  Is there any way to reduce the filesize of these vmdks?

Is there any other method I can use?  Will the OVF be any smaller?  Can it be converted to a VHD?

Sorry for the n00b questions, but I have little / no experience with VMWare, ESX, et al.

Thanks in advance,

Volstag

[continuum]

Datastorebrowser does not display vmdk files correctly as it wrongly assumes that the users are confused by showing the truth.

In reality a VMDK consistes of two files : a short descriptor textfile name.vmdk and a large binary file which keeps the actual data named name-flat.vmdk

Your VM had two virtual disks so you should have name.vmdk,name-flat.vmdk,name_1.vmdk , name_1-flat.vmdk.
If you want a correct view of the files you have use WinSCP or FastSCP

If you copy vmdks with Datastorebrowser to your admin host it will arrive in the nomainal size  - not in the currently used size.
So if you have a vmdk with nominal size of 500 GB and current size of 10 GB and copy it with Datastorebrowser you get a 500Gb file

If you want a smaller size you could use OVF export ...

[a_p_]

Welcome to the Community,

what's the difference between the *.vmdk and the *_1.vmdk?

The virtual machine has two virtual disks. By default the second virtual disk is named "_1".

...  machinename_1.vmdk and machinename_1-flat.vmdk ...

Each virtual disk on an ESXi host consists of two files. The descriptor file (the small text file) and the data file (the -flat.vmdk). The datastore browser does not show both files. You will just see the name of the descriptor file with the size of the flat file.

Now about the disk sizes. It looks like the disks are thin provisioned, which means used disk space on the datastore for the flat file will increase as needed. Since Thin provisioning is a feature of the VMFS file system, the downloaded files will grow to the provisioned virtual disk sizes.

André

[Volstag]

Thanks gents, this helps me understand this a little better.

My concern now is if the *.vmdk provisioned size is greater than 500 GB (as that's the limit of my storage capacity).

Final question (and I can't seem to get a definitive answer Mr. Google): can an OVF be converted to VHD?

Thanks again, really appreciate the input!

-V

[continuum]

The shortest way to a VHD probably is to export the esxi-vmdk as a split-growing vmdk type.
That file format then can be converted to VHD with Wnimage or Starwind Converter

to create that split-growing vmdk you would use the vmkfstools like this:

cd /vmfs/volumes
cd datastore/directory of the vmdk
vmkfstools -i name.vmdk name-export.vmdk -d 2gbsparse

that will create a set of files altogether about the size of the current vmdk

[a_p_]

Just a quick hint about converting to 2GB sparse. There seems to be a flaw in vmkfstools (not sure which versions are affected) where the metadata is not created properly. see Re: vmdk compatibility problem

André