> I have already tried inflate option but it converts to "eagerzeroed thick" which i don't want.
I doubt that you would like the results of an inflate option with a type=lazyzeroed switch.If you have a good use case for such a disk I would like to hear about.
A thin disk references every one-MB block filled with zeroes as a reference to /dev/zero.
To inflate such a disk in place without copying the full disk to a new location (like all gui-clone options and vmkfstools -i do)
ESXi finds an unused block and runs the command
dd if=/dev/zero bs=1M count=1 seek=<offset of unused block>
and then writes a new entry to the fragment-location table of the already existing vmdk.
It does that for every reference to /dev/zero it found for the thin disk.
If the dd command to zerofill the referenced physical disk is skipped you actually fill the vmdk with garbage.
That is not as bad as it may sound - but nothing you may want on purpose.
During recoveries I learned that if I extract lazyzeroed vmdks without using the VMFS-metadata they will come out with a dirty NTFS, often they have invalid MBRs and partiontables.
To mount such a NTFS-vmdk I often have to use Testdisk to fix bootsectors or use the backup MFT because the primary one is broken.
Eagerzerofilled vmdks in most cases come out as clean.
So in my humble opinion we should be glad that your wish can not be fulfilled.
On the other hand ... I would have more customers if your wish came true.
<egoistic mode on>
So lets file a feature request
<egoistic mode off>