Change VM disk from thick EagerZero to thick LazyZero

[ricky73]

I wrong to create hard disk as tick eager zeroed and I just installed my VM, now I need correct it changing hard disk type as tick lazy zeroed. I tried to set lazy disk during datastore vmotion but I can see always eager zeroed. Suggestions?!?

[Rubeck]

Hi...

Any luck if migrating to thin first then to lazy zeroed afterwards as described in this KB?

Eager Zeroed Thick disks cannot be converted back to Thick using Storage vMotion, Cloning, Migration...   

/Rubeck

[Sanjeevan1]

Hi Ricky,

I am not sure why you want change from eager to lazy. Are you doing storage vmotion between two datastore?

Kind regards

Sanjeevan.

[unsichtbare]

There is a misconception that Eagerzeroed is better - it is not.

Eagerzeroed disks have a specific use case, and otherwise present a tremendously larger workload to the SAN, during and after provisioning.

[continuum]

Just curious : what is your use-case that makes you prefer lazyzeroed vmdks ?
Once you have created the VM I dont know why you would want to change to a less stable format:

[unsichtbare]

Well,

Eagerzeroedthick is "required" for MSCS quorum disks and generally recommended for other types of transactional disks.

I use lazy zeroed for: OS, Apps, user share, file servers, etc.

Why would you regard Eagerzeroedthick as a more stable type of disk?

[continuum]

> Why would you regard Eagerzeroedthick as a more stable type of disk?
An eagerzeoed thick vmdk can be carved out of a raw disk and typically does not even require a chkdsk if it was formatted with NTFS.
A lazyzeroed thick vmdk - also using NTFS - in the same conditions typically requires intensive chkdsk clean up and often comes out with damaged areas.
Linux or BSD-filesystems will have even more problems when lazyzeried thick was used.
An eagerzeoed vmdk without the VMFS-metadata is still a clean disk-image.
Lazyzeroed vmdks without the VMFS-metadata are filled with garbage in unpredictable locations.
If you ever have to recover vmdks from damaged VMFS-volumes you will probably regard the longer creation time as peanuts compared with the much better recovery chances.

[ricky73]

Because I use master VM (with eager zeroed disk), which contains apps, to replicate other VM and I suspect to have greater time vm cloning step, which is necessary every time app is modified. Initially my master vm had thin virtual disk because I wrong to create it, I wanted thick lazy zeroed disk! So I run inflate command to vmdk file, then I removed vm from inventory and I added it again; unfortunately virtual disk was eager zeroed and not lazy zeroed, so I tried to convert disk by vmotion phase but it had no results. Disk now are always in eager zeroed format. I ask myself what difference to convert disk to thick by inflate command and vmotion.

[ricky73]

infact I noted to clone vm, time is increased

[ricky73]

>A lazyzeroed thick vmdk - also using NTFS - in the same conditions typically requires intensive chkdsk clean up and often comes out with damaged areas.

>Linux or BSD-filesystems will have even more problems when lazyzeried thick was used.

I think this one depends on storage reliability, e.g. with HP 3par storage clusters are often verified.

I noted for create eager zeroed disk it's necessary more time than lazy zeroed.

I know eager disk are used where there is high I/O traffic to disk

[NagangoudaPatil]

Yes you can change VM disk from thick EagerZero to thick LazyZero by using Standalone converter, select destination disk type and move VM to another host (V2V). I did it for MSCS Cluster requirements, servers are created and later we did V2V and change VM disk type.