Converting Thick to Thin - Stays Thick! need help

[ctstechnology]

Hello,

I have a few VM's that are using "Thick Provision Eager Zeroed" Virtual Disk's that don't need to be, and I am trying to convert them to Thin.

Here is what I have tried so far:

Try #1

Using vSphere client connected to vCenter bundled with vSphere Essentials Plus 5, Shutting down the VM in question -> Right-Click -> Migrate -> Change Datastore -> "Select a virtual disk format: Thin Provision -> Selected a different Datastore

Result: Only Hard Disk 1 gets converted to Thin!, Hard disk 2 stays "Thick Provision Eager Zeroed"

Try #2

Using vSphere client connected to vCenter bundled with vSphere Essentials Plus 5, Shutting down the VM in question -> Right-Click -> Migrate -> Change Datastore -> Advanced -> Selected a different Datastore for each Hard Disk -> Disk Format set to "Thin Provision" for each Hard Disk

Result: Hard disk 2 stays "Thick Provision Eager Zeroed"

Try #3

Removed VM from Inventory and Re-added

Result: Hard disk 2 stays "Thick Provision Eager Zeroed"

Try #4

Enabled SSH on the current ESXi Host in the Cluster ->

Putty to ESXi SSH Command Line -> cd /vmfs/volumes/OS* (Current location of the VMDK files)

-> vmkfstools -i HardDisk1.vmdk -d thin thinHardDisk1.vmdk

-> vmkfstools -E HardDisk1.vmdk thickHardDisk1.vmdk

-> vmkfstools -E thinHardDisk1.vmdk HardDisk1.vmdk

-> vmkfstools -U thickHardDisk1.vmdk

-> removed VM from Inventory and Readded

Result: Hard disk 2 stays "Thick Provision Eager Zeroed"

Note: also tried this using "mv" instead of "vmkfstools -E" with no change in the results

Any Ideas? I really didn't want to have to use the converter just to change the Disk format

Why did Hard Disk 1 get converted fine? It may have been "lazy zeroed" and not "eager zeroed" if that makes a difference.

[jfrappier]

I would suggest creating a test VM and confirming what your setup was.  Hopefully this can be useful:  http://pubs.vmware.com/vsphere-50/index.jsp?topic=%2Fcom.vmware.vsphere.vm_admin.doc_50%2FGUID-4C0F4...

[ctstechnology]

Thank you for your reply.

I have created a few Test VMs. Both with Two Virtual Hard Disks each. One with Lazy zeroed, and the other with Eager zeroed. Migrating to a new datastore and changing the type to thin, was able to successfully convert the virtual disks to thin (on the TEST VMs) without issue.

Here is the setup

Original Setup last year

vSphere Essentials Plus 4.1

x3 HP DL380 G7 Servers (one used for vCenter/Veeam/Management, x2 in a 2-node cluster)

x1 HP P2000 G3 FC Virtualization SAN Starter Kit (x2 Brocade FC Switches/modules/cables, x3 FC Cards, P2000 G3 2.5" SAN)

SAN: HP 600GB 2.5" 10K SAS Drives. Arrays = OS (RAID5), DB (RAID10), LOG (RAID1), DATA (RAID5) and a few global hot spares.

x1 HP SAS DAS for Backups

x2 HP(3com) A5120 Stacked Switches for VM Network/Core switch

Changes made earlier this year:

Moved vCenter to a VM

Moved Veeam to a different physical server

Upgraded to vSphere 5 and added the 3rd DL380 to the Cluster (Making it a 3-node Cluster)

vSphere upgrade was done properly, upgrading the datastores/VM hardware/VMware Tools.

Note that this issue (vSphere ignoring my efforts to convert from thick to thin) is happening on VMs that were created AFTER the upgrade to vSphere 5. So I don't think i can blame upgraded VM drivers or anything like that.

Another note: I have moved the VMDK files to various Datastores and this issue is happending regardless of which Datastores I am migrating to/from.

Let me know if more details would help. Hopefully someone has seen this happen before.

[Niels201110141]

Has this issue been resolved yet? Because I am encountering the exact same issue. Whenever I try to convert them back to Thick Provisioned Lazy Zeroed the system converts it successfully, but the VM will actually report it's Eager Zeroed. This occured to me when I temporarily moved the datastore to a Q-NAP NAS (ext4 formatted, via NFS) and moved them back later from the Q-NAP NAS to the production storage. The server where I moved them from is running VMware vSphere 5.1. The new server is running VMware vSphere 5.0 Update 1 IBM Customized with the latest IBM driver patch.

Things I've tried:

- Upgrading to the latest 5.0 Build

- Using the latest 5.1 Build

- Executing the commands in maintenance mode

- Removing the VM from the inventory and adding it back

- Separating the disks from the VM and then removing the VM from the inventory

- Adding the VM back to the inventory and adding the disks back to the VM.

When you take the last step from the things I've tried, the VM will initially actually tells you it's Lazy Zeroed, but when you press "OK" and "Edit Settings" again it'll just tell you it's Eager Zeroed again.

[RaBalder]

Hi...

I read somewhere that if you have a hardware-accelereated datamover doing the migration of your disk, it will be done fast. But not perhaps do what is expected when converting from a thick to a thin disk.

Have you tried disabling the hardware-accelerated datamover in Advanced Setting on the ESXi and then migrate the data from one datastore to another one?

Without hardware-accelerated datamover, more of the processing will be done by the ESXi and the thick->thin would probably be more like expected.

Also if you move between datastores with different block-sizes, the hardware-accelerated datamover will be skipped because it can handle different block-sizes.

,

[vmjoe]

I have this problem too. The second disk of some VMs will stay on "Thick Provision Eager Zeroed" no matter what I do.

The VM might be an older one though, created as vHW 4 and then upgraded to vHW 7 and now 8. Running on Esxi 5.0.0 build 821926.

In the .vmdk file the parameter ddb.adapterType is set to "legacyESX". Does this maybe prevent it from being converted to thin?

[vmjoe]

The "sdelete -c" trick worked for me on all VMs so far!

In one instance, I had sdelete run and then increased the vmdk by some GB, and before the storage migration to convert it to thin - the disk showed up as thin already!

So it's a bug in VMware I guess.

All my VMs were converted from SAN storage to NAS storage with the thin option, yet a certain amount of machines has "thick eager zeroed" disks (often the secondary 😧 disk).

[vvermani]

I am having a similar problem in my home lab however I n my case it is while creating a new VM using custom settings - I don't have an option to select "Disk Provisioning" which is greyed out with "thin disk" already selected (that is because I am using NFS Datastore).

VM gets created in seconds. When I go back to check VM settings it shows "Thick Provisioned Eager Zeroed".
This happens everytime I create a new VM.

I am running ESXi 5.1 BUILD 799733 (Upgraded from ESXi 5.0U1 which was also upgraded from ESXi 5.0)

On a second note : I don't get "VMware Hardware version 9" as an option while creating a Virtual Machine but I am able to upgrade to, by right clicking the VM itself.