Converting P2V with Thin Provision HD

[Arie01]

Hello,

I'm converting some physical machines to virtual with thin provisioned hard disks but the resultant .vmdk file on the file system is as large as I set it to be, not as needed (or used) by the VM.

When I check the settings of the VM in the vSphere client, it says that the hard drive was thin provisioned.

Why am I getting full sized .vmdk files then?

Thank you,

[Troy_Clavell]

The provisioned size should equal the "thick" size of the vmdk.  What does "used space" look like?

[continuum]

> Why am I getting full sized .vmdk files then?

If your harddisk was used for a long time it may not have zeroed disk space - that applies to block based imports only.
If you use block based cloneing - wipe the unused space with zeroes first

[Arie01]

Hi Troy,

I'm sorry, I don't understand your answer.

I was under the impression that thin provisioned disk is similar to split disk in VMware server, that splits your Hard disk to multiple files of 2GB in size and uses only the space needed by the actual data stored on the drive then grows as needed to accomodate more data. So my data is about 30GB right now and the provisioned disk I coverted to is 300GB. The .vmdk file in the file system is 300GB, not 30GB (or close to it).

[continuum]

You will only get a vmdk close to 30 Gb if you wipe the unused space before the P2V.

Thin provision is something complete different than the sparse split format used with VMserver

[Arie01]

Hi Continuum,

I'm not sure I know what block size cloning is, but I think this is what was done. When I selected Advanced and changed provisioning from thik to thin, there was a yellow triangle with an exclamation mark in it. When I tried changing the block size from 4KB to anything else, it gave me a red warning.

How do I wipe unused space with zeros?

Thank you,

[continuum]

there are several tools to do that - have a look at sdelete.exe for example

A detailed description of tools for this task can be found here http://www.feyrer.de/g4u/#shrinkimg

[a_p_]

Zeroing unused disk space is one way, but shouldn't switching to "volume based" (rather than "disk based") conversion do the job too?

André

[Arie01]

Hi Continuum,

I'm trying to convert another machine with two partitions/drives now and for one it gives me that triangel with the exclamation mark saying (file level) and for the other there is no warning and it says (block level). Is that the right way to do it?

By the way, regarding your comment with the link to g4u, I'm converting servers that have RAID1 or 5 configurations. Does g4u work on servers?

Thank you,

[LuigiC]

you're looking at the vmdk size with du I presume ?

[Arie01]

Hi LuigiC,

What is du?

[LuigiC]

I guess not ... read this

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100541...

[Arie01]

Hi LuigiC,

I read the article you sent me but I still don't know what du is or are.

Also, I'm using ESXi 4.1, this article refers to ESX.

My VMs are currently stored on a NAS before I transfer them to my ESXi server at a remote location.

When I said that the .vmdk file is occupying the full space I set the drive size to, I meant in Windows Explorer, when I browse the NAS share where the VMs are stored. So, if I set a hard drive to 300GB thin provisioning and the data on that physical drive occupying only 30GB, the resultant .vmdk file size is 300GB. That doesn't seem thin to me.

I would like to know what I am doing wrong or what I am missing in my VM conversion settings in order to get a thin .vmdk files.

Mind you, I'm converting production servers using VMware Converter Stand Alone 5.0.0 build-470252.

Thank you,

[Sawkat]

Hi Arie,

During many conversions, we have had similar issues, here is what I did, depending on how long the server had been into production and what type of data were kept, on physical disk, we have fragmentation. So best thing to do is to a defrag first, then after migration if the issue is still there, do a svmotion selecting thin disk, hope you should get better luck there.

[Arie01]

Hi Sawkat,

How do I do svmotion? Remember, I have ESXi with no SSH access. Thank you,

[Sawkat]

how many datastores you have? do you have a vcenter? I guess from what I read you have access to the esxi via the client, so this should be enough.

If you do not have vcenter, then use a free version of the standalone converter. One trick if you have only one datastore, export your vm and reimport it with thin provisioning... I kow its cumbersome, but at least you can get your things right...

to answer your question literally: svmotion is a vmotion of a vm except that instead of changing the host, you migrate only the files from one datastore to another...

[Arie01]

Hi Sawkat,

I have two datastores on the ESXi server and a NAS connected via NFS. I use the vSphere client to access my host server. Is it the same as vcenter?

Thank you,

[Sawkat]

Hey Arie,

If you have 2 DS then move the vms files from one to another or to the nfs (vice versa depending where the files are originally stored). You can use the client to connect to the Host alone or the vcenter. I would suggest you to connect to the VC, since vmotion is only enabled via vcenter. Do you have a vcenter?

[Arie01]

Hi Sawkat,

I have something called VMware vCenter Converter Standalone Client, would that do it?

[Sawkat]

yep, that would very much do